CN114970841A

CN114970841A - Training method of battery state prediction model and related device

Info

Publication number: CN114970841A
Application number: CN202110197414.8A
Authority: CN
Inventors: 胡明睿; 程康
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2021-02-22
Filing date: 2021-02-22
Publication date: 2022-08-30
Also published as: WO2022174601A1

Abstract

The application discloses a training method and a related device for a battery state prediction model in the field of artificial intelligence. According to the technical scheme, a first model is subjected to self-supervision training according to first battery characteristic information of a target battery, and the first model comprises a first pre-training representation model and a first prediction model; and performing supervision training on the battery state prediction model according to second battery characteristic information and first label information of the target battery, wherein the battery state prediction model comprises a characteristic vector extraction model and a second prediction model, and initial parameters before the characteristic vector extraction model performs supervision training comprise parameters of a first pre-training representation model obtained after the first model is subjected to self-supervision training. According to the technical scheme, the battery state of the real vehicle can be predicted by using the real vehicle data, and the training cost of the battery state prediction model is reduced.

Description

Training method of battery state prediction model and related device

Technical Field

The present application relates to the field of artificial intelligence, and in particular, to a training method for a battery state prediction model and a related apparatus.

Background

Artificial Intelligence (AI) is a theory, method, technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results. In other words, artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a manner similar to human intelligence. Artificial intelligence is the research of the design principle and the realization method of various intelligent machines, so that the machines have the functions of perception, reasoning and decision making. Research in the field of artificial intelligence includes robotics, natural language processing, computer vision, decision and reasoning, human-computer interaction, recommendation and search, AI basic theory, and the like.

Electric Vehicles (EVs) have become the main direction of the development of modern vehicles due to their advantages of low pollution and high performance. Meanwhile, the battery state prediction of the electric automobile also becomes a focus of the industry.

At present, models for predicting various states of a battery (for example, SOC information, SOH information, failure information, or remaining life information) are end-to-end prediction models, and are trained from a zero initial state, and thus the data volume requirement on tag data is extremely high.

Disclosure of Invention

The application provides a training method and a related device of a battery state prediction model, which can predict the battery state of an EV (electric vehicle) by using real vehicle data and reduce the training cost of the battery state prediction model.

In a first aspect, the present application provides a training method for a battery state prediction model, where the training method includes: acquiring first battery characteristic information of a target battery in a first time period, wherein the battery characteristic information of the target battery comprises one or more of the following information: the current information of the target battery, the voltage information of the target battery, the temperature information of the target battery and the battery power index SOC of the target battery are obtained; performing self-supervision training on a first model according to the first battery characteristic information, wherein the first model comprises a first pre-training representation model and a first prediction model, the input of the first prediction model comprises the output of the first pre-training representation model, the first pre-training representation model is used for determining a representation vector of the input battery characteristic information, and the first prediction model is used for determining target battery characteristic information corresponding to the input representation vector; acquiring second battery characteristic information and first label information of the target battery, wherein the first label information is used for indicating battery state information corresponding to the second battery characteristic information; and performing supervision training on the battery state prediction model according to the second battery characteristic information and the first label information, wherein the battery state prediction model comprises a characteristic vector extraction model and a second prediction model, the input of the second prediction model comprises the output of the characteristic vector extraction model, and the initial parameters of the characteristic vector extraction model comprise the parameters of the first pre-training expression model obtained after the self-supervision training is performed on the first model.

In the method, after the label-free data is used for carrying out the self-supervision training on the first model, the association knowledge between the battery characteristic information and the battery state information can be learned in the parameters of the pre-training representation model obtained through training, so when the initialization parameters of the labeled data comprise the battery state prediction model trained by the pre-training representation model, the requirement of the battery state prediction model on the labeled data can be reduced. In other words, the battery state prediction model can also be trained using a small amount or stock of labeled data.

Optionally, each battery characteristic information in the method may be a battery raw data characteristic, for example, a battery raw data characteristic acquired via a sensor. For example, the battery characteristic information may include current information, voltage information, temperature information, SOC information, and the like. Therefore, the battery state model obtained through training can be used for predicting the state directly according to the battery original data, namely, the battery state can be predicted based on the real-time data of the battery.

With reference to the first aspect, in a first possible implementation manner, a structure of the second prediction model is the same as that of the first prediction model, and initial parameters of the second prediction model before the supervised training is performed include parameters of the first prediction model obtained after the unsupervised training is performed on the first model.

In this implementation, the structure of the second prediction model is the same as the structure of the first prediction model, and the initial parameters of the second prediction model include the parameters of the first prediction model after the completion of the self-supervision training. The first prediction model after the self-supervision training has learned the incidence relation between the expression vector of the battery characteristic information and the battery characteristic information, which is equivalent to that the second prediction model is pre-trained before being trained, so that the training efficiency of the battery state prediction model is improved.

With reference to the first aspect or the first possible implementation manner, in a second possible implementation manner, the first battery characteristic information includes N battery characteristic information of the target battery at N times within the first time period, where N is a positive integer, the first pre-training representation model includes a first encoder and a second encoder, an input of the first encoder includes the battery characteristic information, and an input of the second encoder includes an output of the first encoder.

With reference to the second possible implementation manner, in a third possible implementation manner, the performing self-supervision training on the first model according to the first battery characteristic information includes: dividing the N pieces of battery characteristic information into N subsets, wherein each subset of the N subsets comprises at least one piece of battery characteristic information, the time corresponding to any battery characteristic information in the ith subset of the N subsets in the first time is earlier than the time corresponding to any battery characteristic information in the (i + 1) th subset of the N subsets in the first time, N is a positive integer and N is less than or equal to N, i is a positive integer and i is less than N; inputting a battery characteristic information sequence obtained by arranging all the battery characteristic information in each subset of the n subsets according to the corresponding time sequence into the first encoder to obtain a representation vector corresponding to each subset; inputting a battery characteristic information sequence obtained after arranging the n expression vectors corresponding to the n subsets according to a corresponding time sequence or a sequence from low SOC value to high SOC value into the second encoder to obtain a first expression vector of the first battery characteristic information; inputting the first expression vector into the first prediction model to obtain target battery characteristic information corresponding to the first expression vector; and adjusting parameters of the first model according to the target battery characteristic information and the first battery characteristic information.

In the implementation mode, N pieces of battery characteristic information in a first time period are divided into N subsets, all battery characteristic information in each subset of the N subsets is arranged according to corresponding time sequence to obtain a battery characteristic information sequence, and the battery characteristic information sequence is input into a first encoder to obtain a representation vector corresponding to each subset; sequentially inputting the expression vectors corresponding to each subset into a second encoder according to the time sequence or the sequence of the SOC values from low to high to obtain a first expression vector of the first battery characteristic information; inputting the first expression vector into a first prediction model to obtain target battery characteristic information corresponding to the first expression vector; and parameters of the first model are trained according to the target battery characteristic information and the first battery characteristic information, so that the accuracy of the first model is improved, and the training cost is reduced.

With reference to the third possible implementation manner, in a fourth possible implementation manner, the dividing the N battery characteristic information into N subsets includes: and dividing the battery characteristic values into n subsets according to the SOC values, wherein the SOC values corresponding to all battery characteristic information contained in each of the n subsets are the same.

With reference to the second, third, or fourth possible implementation manner, in a fifth possible implementation manner, the target battery is a battery pack, and the first battery characteristic information and/or the second battery characteristic information further include voltage information of each single battery cell in the battery pack.

With reference to the first aspect, in a sixth possible implementation manner, the first battery feature information includes M battery feature information of the target battery at M times in a first period, where M is a positive integer, the first pre-training representation model includes a variational encoder, the first prediction model includes a variational decoder, and the second prediction model includes a regression model or a classification model.

With reference to the sixth possible implementation manner, in a seventh possible implementation manner, the performing self-supervision training on the first model according to the first battery characteristic information includes: dividing the M pieces of battery characteristic information into M subsets, wherein each subset in the M subsets comprises at least one piece of battery characteristic information, the time corresponding to any battery characteristic information in the jth subset in the M subsets in the first time is earlier than the time corresponding to any battery characteristic information in the (j + 1) th subset in the M subsets in the first time, M is a positive integer and M is less than or equal to M, j is a positive integer and j is less than M; determining a representation vector corresponding to each subset according to all battery characteristic information in each subset of the m subsets; calculating the average value of m expression vectors corresponding to the m subsets; inputting the average value into the first pre-training representation model to obtain a second representation vector of the first battery characteristic information; inputting the second expression vector into the first prediction model to obtain target battery characteristic information corresponding to the second expression vector; and adjusting parameters of the first model according to the target battery characteristic information and the first battery characteristic information.

In the implementation mode, dividing M pieces of battery characteristic information in a first time period into M subsets, and determining a representation vector corresponding to each subset according to all the battery characteristic information in each subset in the M subsets; calculating the average value of m expression vectors corresponding to the m subsets; inputting the average value into a first pre-training representation model to obtain a second representation vector of the first battery characteristic information; inputting the second expression vector into the first prediction model to obtain target battery characteristic information corresponding to the second expression vector; and parameters of the first model are trained according to the target battery characteristic information and the first battery characteristic information, so that the accuracy of the first model is improved, and the training cost is reduced.

With reference to the seventh possible implementation manner, in an eighth possible implementation manner, the number of the battery characteristic information in any one of the m subsets is equal to the number of the battery characteristic information in any other one of the m subsets.

With reference to the sixth possible implementation manner, the seventh possible implementation manner or the eighth possible implementation manner, in a ninth possible implementation manner, the target battery is a single battery cell, and the first battery characteristic information and/or the second battery characteristic information includes insulation resistance information of the single battery cell and/or an electrochemical alternating current impedance spectrum EIS of the single battery cell.

With reference to the first aspect or any one of the foregoing possible implementation manners, in a tenth possible implementation manner, the target battery characteristic information includes: the SOC sequence of the target battery in the first time period, the voltage sequence of the target battery in the first time period, the charging time length sequence of each cell in the target battery in the first time period, the charging mode of the target battery, or the SOC value of the target battery at a random mask in the first time, where the charging mode includes a fast charging mode or a slow charging mode.

With reference to the first aspect or any one of the foregoing possible implementation manners, in an eleventh possible implementation manner, the battery state information includes battery health indicator SOH information or fault information or remaining life information.

In a second aspect, the present application provides a method for predicting a battery state, the method comprising: acquiring characteristic information of a battery to be predicted of the battery to be predicted; and determining the battery state information of the battery to be predicted based on the characteristic information of the battery to be predicted by using a battery state prediction model, wherein the battery state prediction model is obtained by training by using the training method in the first aspect or any one of the possible implementation manners.

In the method, the battery characteristic information to be predicted of the battery to be predicted can be directly obtained from data reported by the EV, so that the EV can predict the battery state according to real vehicle data.

In a third aspect, the present application provides a training apparatus for a battery state prediction model, which may include various modules for implementing the method in the first aspect, and these modules may be implemented by software and/or hardware.

In a fourth aspect, the present application provides a device for predicting battery status, which may include various modules for implementing the method in the second aspect, and the modules may be implemented by software and/or hardware.

In a fifth aspect, the present application provides a training apparatus for a battery state prediction model. The apparatus may include a processor coupled with a memory. Wherein the memory is configured to store program code and the processor is configured to execute the program code in the memory to implement the method of the first aspect or any one of the implementations.

Optionally, the apparatus may further comprise the memory.

In a sixth aspect, the present application provides a battery state prediction apparatus. The apparatus may include a processor coupled with a memory. Wherein the memory is configured to store program code and the processor is configured to execute the program code in the memory to implement the method of the second aspect.

Optionally, the apparatus may further comprise the memory.

In a seventh aspect, the present application provides a chip, which includes at least one processor and a communication interface, where the communication interface and the at least one processor are interconnected by a line, and the at least one processor is configured to execute a computer program or instructions to perform the method according to the first aspect or the second aspect or any one of the possible implementation manners.

In an eighth aspect, the present application provides a computer readable medium storing program code for execution by a device, the program code comprising instructions for performing a method according to the first or second aspect or any one of its possible implementations.

In a ninth aspect, the present application provides a computer program product comprising instructions which, when run on a computer, cause the computer to perform the method according to the first aspect or the second aspect or any one of its possible implementations.

In a tenth aspect, the present application provides a computing device comprising at least one processor and a communication interface, the communication interface and the at least one processor being interconnected by a line, the communication interface being in communication with a target system, the at least one processor being configured to execute a computer program or instructions to perform a method according to the first aspect or the second aspect or any one of its possible implementations.

In an eleventh aspect, the present application provides a computing system comprising at least one processor and a communication interface, the communication interface and the at least one processor being interconnected by a line, the communication interface being in communication with a target system, the at least one processor being configured to execute a computer program or instructions to perform a method according to the first aspect or the second aspect or any one of its possible implementations.

According to the training method of the battery state prediction model, an embedded representation (embedding representation) idea and a pre-training assembly line idea in deep learning are introduced, the state of the battery in any detection section is represented as an abstract battery representation vector, label-free data reported by a battery pack is fully utilized, and the battery state prediction model is trained through at least one pre-training task designed according to the electrochemical characteristics of the battery.

For the trained battery state prediction model, a transfer learning method can be used for executing subsequent end-to-end data driving tasks of all batteries, a pre-trained representation model in the battery state prediction model can be used for directly extracting representation vectors of the batteries for the subsequent battery state prediction task, dependence on a large amount of label data needing manual labeling is reduced, prediction of the battery state of the EV only depends on general data reported by the EV, feature transformation of the general data reported by the EV is not needed, hardware of the EV is not needed to be modified, and cost is reduced.

The training method of the battery state prediction model can be adapted to EV battery packs of different manufacturers and different structures or single battery cores of different material systems through transfer learning.

According to the pre-training representation model in the battery state prediction model, a transformer encoder model can be used, a two-layer inheritance encoder model is designed, samples of each time node are fully utilized, a time sequence with variable length can be processed, and the calculation amount is small.

When the real vehicle data volume reported by the EV is large enough, the method can cover various characteristics of voltage, temperature and current, and can fully excavate the nonlinear relation between input data by enlarging the scale of a battery state prediction model, thereby achieving the effect of covering complex working conditions and the internal structure of a complex battery pack.

Drawings

Fig. 1 is a schematic diagram of a system architecture provided in an embodiment of the present application;

fig. 2 is a schematic structural diagram of chip hardware according to an embodiment of the present disclosure;

FIG. 3 is a schematic diagram of another system architecture provided by embodiments of the present application;

fig. 4 is a flowchart illustrating a training method of a battery state prediction model according to an embodiment of the present application;

FIG. 5 is a flowchart illustrating a method for training a SOH prediction model of a battery pack according to an embodiment of the present application;

FIG. 6 is a schematic flow chart illustrating a method for training a fault prediction model of a battery pack according to an embodiment of the present application;

fig. 7 is a schematic flowchart of a training method of an SOH prediction model of a cell unit according to an embodiment of the present application;

fig. 8 is a schematic flowchart of a method for training a failure prediction model of a cell in accordance with an embodiment of the present application;

fig. 9 is a schematic flowchart of a battery state prediction method according to an embodiment of the present application;

FIG. 10 is a schematic block diagram of a training apparatus for a battery state prediction model according to an embodiment of the present application;

fig. 11 is a schematic configuration diagram of a battery state prediction apparatus according to an embodiment of the present application;

fig. 12 is a schematic structural diagram of an apparatus according to an embodiment of the present application.

Detailed Description

Technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

Fig. 1 is a schematic diagram of a system architecture according to an embodiment of the present application. Referring to fig. 1, the data collecting device 160 is configured to collect battery characteristic information of a target battery and battery status label information corresponding to a part of battery characteristics, and store the battery characteristic information and the battery status label information in the database 130, where the battery characteristic information without corresponding to the battery status label information is referred to as first battery characteristic information, the battery characteristic information with corresponding to the battery status label information is referred to as second battery characteristic information, and the battery status label information corresponding to the second battery characteristic information is referred to as first label information; the training device 120 generates a battery state prediction model 101 based on the first battery characteristic information, the second battery characteristic information, and the first label information maintained in the database 130, where the battery state model may also be referred to as a battery state prediction rule.

A method for the training apparatus 120 to obtain the battery state prediction model 101 based on the first battery characteristic information, the second battery characteristic information and the first label information may refer to an embodiment illustrated in any of fig. 4 to 8.

The battery state prediction model 101 obtained by the training device 120 may be applied to different systems or devices, such as the execution device 110.

The execution device 110 is configured with an I/O interface 112 for data interaction with an external device, and a "user" can input battery characteristic information to be predicted of a battery to be predicted to the I/O interface 112 through the client device 140.

The execution device 110 may call data, code, etc. from the data storage system 150 and may store data, instructions, etc. in the data storage system 150.

The calculation module 111 uses the battery state prediction model 101 to process the battery characteristic information of the battery to be predicted, so as to obtain the battery state information of the battery to be predicted.

Finally, the I/O interface 112 returns the results of the processing to the client device 140 for presentation to the user.

In the case shown in fig. 1, the user may manually specify the battery characteristic information to be predicted in the input execution device 110, for example, operating in an interface provided by the I/O interface 112. Alternatively, the client device 140 may automatically input the information about the characteristics of the battery to be predicted to the I/O interface 112 and obtain the information about the status of the battery, and if the client device 140 automatically inputs the information about the characteristics of the battery to be predicted and requires authorization of the user, the user may set corresponding rights in the client device 140. The user can view the battery status information output by the execution device 110 at the client device 140, and the specific presentation form may be a display, a sound, an action, and the like. The client device 140 may also act as a data collection site to store the collected battery characteristic information and tag information in the database 130.

It should be noted that fig. 1 is only a schematic diagram of a system architecture provided in an embodiment of the present application, and the position relationship between the devices, modules, and the like shown in the diagram does not constitute any limitation, for example, in fig. 1, the data storage system 150 is an external memory with respect to the execution device 110, and in other cases, the data storage system 150 may also be disposed in the execution device 110.

Fig. 2 is a schematic structural diagram of chip hardware according to an embodiment of the present disclosure. Referring to fig. 2, a neural-Network Processing Unit (NPU) is mounted as a coprocessor on a Host central processing unit (Host CPU), and the Host CPU allocates tasks. The core portion of the NPU is an arithmetic circuit 20, and the controller 204 controls the arithmetic circuit 203 to extract data in the memories (the weight memory 202 and/or the input memory 201) and perform arithmetic.

In some implementations, the arithmetic circuitry 203 includes a plurality of processing units (PEs) therein. In some implementations, the operational circuitry 203 is a two-dimensional systolic array. The arithmetic circuitry 203 may also be a one-dimensional systolic array or other electronic circuitry capable of performing mathematical operations such as multiplication or addition. In some implementations, the arithmetic circuitry 203 is a general-purpose matrix processor.

For example, assume that there is an input matrix A, a weight matrix B, and an output matrix C. The arithmetic circuit fetches the data corresponding to the matrix B from the weight memory 202 and buffers it on each PE in the arithmetic circuit 203. The arithmetic circuit takes the matrix a data from the input memory 201 and performs matrix operation with the matrix B, and partial or final results of the obtained matrix are stored in an accumulator (accumulator) 208.

The vector calculation unit 207 may further process the output of the arithmetic circuit, such as vector multiplication, vector addition, exponential operation, logarithmic operation, magnitude comparison, and the like. For example, the vector calculation unit 207 may be used for network calculations of non-convolution/non-FC layers in a neural network, such as Pooling (Pooling), Batch Normalization (Batch Normalization), Local Response Normalization (Local Response Normalization), and the like.

In some implementations, the vector calculation unit 207 may store the processed output vector to the unified buffer 206. For example, the vector calculation unit 207 may apply a non-linear function to the output of the arithmetic circuit 203, such as a vector of accumulated values, to generate the activation value. In some implementations, the vector calculation unit 207 may generate normalized values, combined values, or both. In some implementations, the vector of processed outputs can be used as activation inputs for the arithmetic circuitry 203, e.g., for use in subsequent layers in a neural network.

The unified memory 206 is used to store input data as well as output data.

A memory cell access controller (DMAC) 205 transfers input data in the external memory to the input memory 201 and/or the unified memory 206, stores weight data in the external memory into the weight memory 202, and stores data in the unified memory 206 into the external memory.

A Bus Interface Unit (BIU) 210, configured to implement interaction between the main CPU, the DMAC, and the instruction fetch memory 209 through a bus.

An instruction fetch buffer (issue fetch buffer)209 coupled to the controller 204 is used to store instructions used by the controller 204.

The controller 204 is configured to call the instruction cached in the instruction fetch memory 209 to implement controlling of the operation process of the operation accelerator. Unified memory 206, input memory 201, weight memory 202, and instruction fetch memory 209 are all On-Chip (On-Chip) memories. The external memory is private to the NPU hardware architecture.

In some implementations, the chip shown in fig. 2 may implement the method shown in any of fig. 4-8 to obtain the battery state prediction model. As an example, the method may be performed by the host CPU and the NPU in cooperation.

In other implementations, the chip shown in fig. 2 may implement the method shown in fig. 9 to obtain the battery state information of the battery to be predicted. As an example, the method may be performed by the host CPU and the NPU in cooperation.

Fig. 3 is a schematic diagram of another system architecture provided in an embodiment of the present application. Referring to fig. 3, computing device 310 is implemented by one or more servers and, optionally, may cooperate with other computing devices, e.g., computing device 310 may cooperate with data storage, routers, and load balancers; the computing device 310 may be disposed on one physical site or distributed across multiple physical sites.

In the system shown in fig. 3, computing device 310 may use data in data storage system 350, or call program code in data storage system 150, to implement the method shown in any of fig. 4-8. As one example, computing device 310 may be training device 120 in fig. 1.

Alternatively, in the system shown in FIG. 3, computing device 310 may use data in data storage system 350 or call program code in data storage system 150 to implement the method shown in FIG. 9.

The user may operate respective user devices (e.g., local device 301 and local device 302) to interact with computing device 310. Each local device may represent any computing device, such as a personal computer, computer workstation, smartphone, tablet, smart camera, smart car or other type of cellular phone, media consumption device, wearable device, set-top box, game console, and so forth.

Each user's local device may interact with the computing device 310 via a communication network of any communication mechanism or communication standard, such as a wide area network, a local area network, a peer-to-peer connection, and the like, or any combination thereof.

In another implementation, one or more aspects of the computing device 310 may be implemented by each local device, e.g., the local device 301 may provide local data or feedback calculations for the computing device 310.

It should be noted that all of the functionality of the computing device 310 may also be implemented by a local device. For example, the local device 301 implements the functionality of the computing device 310 and provides services to its own user or provides services to a user of the local device 302.

Fig. 4 is a flowchart illustrating a method for training a battery state prediction model according to an embodiment of the present disclosure, and as shown in fig. 4, the method at least includes S401 to S404.

S401, first battery characteristic information of the target battery in a first time interval is obtained.

The target battery may be a battery pack including a plurality of unit cells, or may be a unit cell. The first period may be any period, and the duration of the first period may be predetermined.

It is understood that the first battery characteristic information of the target battery during the first period may include battery characteristic information of the target battery at each of at least one time point during the first period.

As an example, when the target battery is a battery pack, the first battery characteristic information may include current information of the battery pack, voltage information of the battery pack, temperature information of the battery pack, voltage information of each cell in the battery pack, and a battery state of charge (SOC) of the battery pack.

As another example, when the target battery is a single cell, the first battery characteristic information may include voltage information of the single cell, temperature information of the single cell, current information of the single cell, SOC information of the single cell, insulation resistance information of the single cell, or Electrochemical Impedance Spectroscopy (EIS) of the single cell.

In some implementations, the first battery characteristic information of the target battery may be directly obtained from data reported by a Battery Management System (BMS) of an EV to which the target battery belongs.

S402, performing self-supervision training on a first model according to the first battery characteristic information, wherein the first model comprises a first pre-training expression model and a first prediction model, the input of the first prediction model comprises the output of the first pre-training expression model, the first pre-training expression model is used for determining the expression vector of the input battery characteristic information, and the first prediction model is used for determining the target battery characteristic information corresponding to the input expression vector.

In some implementations, the first pre-trained representation model includes a first encoder and a second encoder, an input of the first encoder includes the first battery characteristic information, an input of the second encoder includes an output of the first encoder, and the first prediction model includes a regressor or a classifier, or the like.

In other implementations, the first pre-training representation model includes a variational encoder (VAE) and the first prediction model includes a variational decoder.

When the first pre-trained representation model includes a two-stage encoder, the first encoder and the second encoder in the first pre-trained representation model may be, for example, a multi-layer inheritance converter (transformer) structure.

In some implementations of this embodiment, the above-mentioned self-supervised training may include at least one of the following training tasks: predicting the SOC sequence of the target battery in a first period, namely the characteristic information of the target battery is the SOC sequence of the target battery in the first period; predicting a voltage sequence of the target battery in a first period, namely the characteristic information of the target battery is the voltage sequence of the target battery in the first period; predicting the charging time of each SOC segment of the target battery in a first period, namely the target battery characteristic information is the charging time of each SOC segment of the target battery in the first period; predicting a charging mode of the target battery in a first time period, namely whether the charging mode of the target battery in the first time period is a fast charging mode or a slow charging mode is the target battery characteristic information; or predicting the SOC value of the target battery at the random mask in the first period, that is, the target battery characteristic information is the SOC value of the target battery at the random mask in the first period.

In this embodiment, one implementation manner of performing the self-supervision training on the first model according to the first battery characteristic information may include steps 1 to 5. In this implementation, the first battery characteristic information includes N battery characteristic information of the target battery at N time points within the first period, where N is a positive integer.

Step 1, dividing N pieces of battery characteristic information in a first time period into N subsets, wherein each subset in the N subsets comprises at least one piece of battery characteristic information, the time corresponding to any battery characteristic information in the ith subset in the N subsets in the first time is earlier than the time corresponding to any battery characteristic information in the (i + 1) th subset in the N subsets in the first time, N is a positive integer and N is less than or equal to N, i is a positive integer and i is less than N.

As an example, the N battery characteristic information may be divided into N subsets according to SOC values of the N battery characteristic information in the first time period, and SOC values corresponding to all battery characteristic information included in each of the N subsets after division are the same.

As another example, the N battery characteristic information may be divided into N subsets according to voltage values of the N battery characteristic information in the first time period, and voltage values corresponding to all battery characteristic information included in each of the N subsets after division are the same.

And 2, inputting a battery characteristic information sequence obtained by arranging all the battery characteristic information in each subset of the n subsets according to the corresponding time sequence into a first encoder to obtain a representation vector corresponding to each subset.

And 3, inputting a battery characteristic information sequence obtained after arranging the n expression vectors corresponding to the n subsets according to the corresponding time sequence or the sequence from low SOC value to high SOC value into a second encoder to obtain a first expression vector of the first battery characteristic information.

And 4, inputting the first expression vector into the first prediction model to obtain target battery characteristic information corresponding to the first expression vector.

And 5, training parameters of the first model according to the target battery characteristic information and the first battery characteristic information.

In this embodiment, another implementation manner of performing the self-supervision training on the first model according to the first battery characteristic information may include steps 6 to 11. The first battery characteristic information may include M battery characteristic information of the target battery at M times within the first period, where M is a positive integer.

Step 6, dividing the M pieces of battery characteristic information in the first time period into M subsets, wherein each subset in the M subsets comprises at least one piece of battery characteristic information, the time corresponding to any battery characteristic information in the jth subset in the M subsets in the first time is earlier than the time corresponding to any battery characteristic information in the (j + 1) th subset in the M subsets in the first time, M is a positive integer and M is less than or equal to M, j is a positive integer and j is less than M.

As an example, the number of the battery characteristic information in each of M subsets may be preset, and the number of the battery characteristic information in any one of the M subsets is equal to the number of the battery characteristic information in any other one of the M subsets, the M battery characteristic information in the first time period is divided according to the preset number of the battery characteristic information in each subset, and if the number of the battery characteristic information in any one of the M subsets does not reach the preset number of the battery characteristic information in each subset, the lacking battery characteristic information is marked as 0.

And 7, determining a representation vector corresponding to each subset according to all the battery characteristic information in each subset of the m subsets.

And 8, calculating the average value of m expression vectors corresponding to the m subsets.

And 9, inputting the average value into the first pre-training representation model to obtain a second representation vector of the first battery characteristic information.

And step 10, inputting the second expression vector into the first prediction model to obtain target battery characteristic information corresponding to the second expression vector.

And 11, training parameters of the first model according to the target battery characteristic information and the first battery characteristic information.

And S403, acquiring second battery characteristic information and first label information, wherein the first label information is used for indicating battery state information corresponding to the second battery characteristic information.

As an example, the target battery may be a battery pack, and the second battery characteristic information may include current information of the battery pack, voltage information of the battery pack, temperature information of the battery pack, voltage information of each cell in the battery pack, an SOC value of the battery pack, and the like.

As another example, the target battery may be a cell, and the second battery characteristic information may include voltage information of the cell, temperature information of the cell, current information of the cell, SOC information of the cell, insulation resistance information of the cell, or EIS of the cell.

In this embodiment, the second battery characteristic information and the first battery characteristic information may be the same battery characteristic information or different battery characteristic information.

In some implementations of this embodiment, the battery state information corresponding to the second battery characteristic information may include SOH information, fault information, or remaining life information of the target battery.

In this embodiment, the second battery characteristic information and the first label information may be collectively referred to as training data.

S404, performing supervision training on the battery state prediction model according to second battery characteristic information and first label information, wherein the battery state prediction model comprises a characteristic vector extraction model and a second prediction model, the input of the second prediction model comprises the output of the characteristic vector extraction model, and the initial parameters of the characteristic vector extraction model before supervision training comprise the parameters of a first pre-training expression model obtained after self-supervision training on the first model.

In some implementation manners, all or part of parameters in a feature vector extraction model in the battery state prediction model may be initialized by using model parameters of a first pre-training representation model obtained after self-supervision training is performed on a first model, and then the feature vector extraction model battery state prediction model is further trained by using second battery feature information and first label information to obtain the battery state prediction model.

In other possible implementation manners, a first pre-training representation model obtained after the self-supervision training of the first model may be used as a feature vector extraction model in the battery state prediction model, second battery feature information is input into the feature vector extraction model to obtain a representation vector of the second battery feature information, and the second prediction model is trained by using the second battery feature information, the representation vector of the second battery feature information, and the first label information to obtain the battery state prediction model.

In this embodiment, when the battery state information includes SOH information, the battery state prediction model may be referred to as an SOH prediction model; when the battery state information includes fault information, the battery state prediction model may be referred to as a fault prediction model.

According to the technical scheme, the first battery characteristic information for performing the self-supervision training on the first model and the second battery characteristic information for performing the supervision training on the second model can be directly obtained from data reported by the EV, the number of the first label information used for performing the supervision training on the second model is greatly reduced compared with the number of the label data used for training the battery state prediction model in the prior art, and the training cost of the battery state prediction model is reduced.

Since the pre-training representation model of the battery state prediction model is trained only using real vehicle data during the training of the battery state prediction model, the battery state of the EV can be predicted only using real vehicle data when the battery state of the EV is predicted using the battery state prediction model.

An exemplary training method of the battery state prediction model is introduced below by taking a target battery as a battery pack, a first pre-training representation model comprising an SOC encoder and an SOC sequence encoder, a first prediction model being a classifier or a regressor, and a battery state prediction model being an SOH prediction model.

Fig. 5 is a flowchart illustrating a method for training a SOH prediction model of a battery pack according to an embodiment of the present application. As shown in fig. 5, the method includes at least S501 to S508.

S501, acquiring all first battery characteristic information of the battery pack in a first detection time length.

Optionally, the first battery characteristic information of the battery pack in the first detection duration may include current information of the battery pack, voltage information of the battery pack, temperature information of the battery pack, voltage information of each single battery cell in the battery pack, SOC information of the battery pack, and the like.

The first battery characteristic information of the battery pack can be obtained based on data reported by an original BMS of the EV.

S502, dividing the first detection time into K time segments according to the SOC value of the battery pack, wherein the SOC value of the first battery characteristic information in each time segment is the same, and K is a positive integer.

The first detection time length is divided into K time segments according to the SOC value of the battery pack in the first detection time length, each time segment corresponds to one SOC value, the SOC values of the first battery characteristic information in each time segment are the same, each time segment at least comprises one first battery characteristic information, and the time segments can also be called SOC frames.

The time corresponding to any first battery characteristic information in the x-th time segment of the K time segments in the first detection time length is earlier than the time corresponding to any first battery characteristic information in the x + 1-th time segment of the K time segments in the first detection time length, x is a positive integer and is smaller than K.

And S503, inputting the first battery characteristic information in each of the K time slices into the SOC encoder to obtain a corresponding expression vector of the time slice.

Each time slice in the K time slices corresponds to a representation vector, and the representation vector is used for representing the first battery characteristic information of the corresponding time slice.

The K time segments of the first detection duration correspond to the K representation vectors.

S504, sequentially inputting the expression vector corresponding to each time slice in the K time slices into the SOC sequence encoder according to the time sequence to obtain the expression vector corresponding to the first detection duration.

The SOC sequence encoder obtains a representation vector corresponding to the first detection duration according to the representation vectors of the K time segments in the first detection duration, and the representation vector corresponding to the first detection duration is used for representing battery feature information of the battery pack in the first detection duration.

And S505, inputting the expression vector corresponding to the first detection duration into a regressor or a classifier to obtain a prediction output result of the battery pack in the first detection duration.

S506, performing self-supervision training on the first pre-training representation model and the first prediction model by using a pre-training task, wherein the first pre-training representation model comprises an SOC encoder and an SOC sequence encoder, and the first prediction model comprises a regressor or a classifier.

Optionally, a plurality of pre-training tasks may be designed to perform the self-supervision training, where the designed pre-training tasks may include predicting an SOC sequence of the battery pack in the first detection duration, predicting a voltage sequence of the battery pack in the first detection duration, predicting a charging time of each SOC segment of the battery pack in the first detection duration, predicting a charging mode of the battery pack in the first detection duration, predicting an SOC value of the battery pack at a random mask in the first detection duration, and the like.

It is understood that when performing the self-supervised training based on a plurality of pre-training tasks, different pre-training tasks may correspond to the same pre-training representation model and different first prediction models. Or, under the condition that a plurality of pre-training tasks are designed, a corresponding first prediction model may be designed for each pre-training task, and then, the same pre-training task and the first prediction model corresponding to the pre-training task are subjected to self-supervision training for each pre-training task in sequence.

For example, when 5 pre-training tasks are designed, a corresponding first prediction model may be designed for each pre-training task, that is, 5 first prediction models are finally designed; then, a first pre-training task in the 5 pre-training tasks is used for training a model formed by a pre-training representation model and a first prediction model corresponding to the first pre-training task; then, a second pre-training task is used for carrying out self-supervision training on a model formed by the trained pre-training representation model and a first prediction model corresponding to the second pre-training task; and repeating the steps until the last pre-training task is used for performing self-supervision training on a model formed by the pre-training representation model obtained by the fourth self-supervision training and the first prediction model corresponding to the last training task.

And S507, acquiring second battery characteristic information of the battery pack and SOH label information corresponding to the second battery characteristic information.

Optionally, the second battery characteristic information of the battery pack may include current information of the battery pack, voltage information of the battery pack, temperature information of the battery pack, voltage information of each single electric core in the battery pack, SOC information of the battery pack, and the like.

And S508, performing supervision training on a battery state prediction model according to second battery characteristic information of the battery pack and SOH label information corresponding to the second battery characteristic information, wherein the battery state prediction model comprises a characteristic vector extraction model and a second prediction model.

In some possible implementation manners, parameters in a feature vector extraction model in the battery state prediction model are initialized by using model parameters of a first pre-training representation model obtained after self-supervision training, and then the feature vector extraction model and the second prediction model are further trained by using second battery feature information and SOH label information corresponding to the second battery feature information to obtain the battery state prediction model.

In other possible implementation manners, the first pre-training representation model obtained after the self-supervision training is used as a feature vector extraction model in the battery state prediction model, the second battery feature information is input into the feature vector extraction model to obtain a representation vector of the second battery feature information, and the second prediction model is trained by using the second battery feature information, the representation vector of the second battery feature information and the SOH label information corresponding to the second battery feature information to obtain the battery state prediction model.

The battery state prediction model obtained according to the training method shown in fig. 5 may be used to predict SOH information of the battery pack.

In the technical scheme provided by the application, all first battery characteristic information of a battery pack in a first detection time length is divided into K time segments according to an SOC value of the battery pack, the battery characteristic information in each time segment of the K time segments is input into an SOC encoder in a first pre-training representation model to obtain a representation vector corresponding to each time segment, the representation vector corresponding to each time segment of the K time segments is sequentially input into an SOC sequence encoder in the first pre-training representation model according to a time sequence to obtain a representation vector corresponding to the first detection time length, the representation vector corresponding to the first detection time length is input into a first prediction model to obtain a prediction output result of the battery pack in the first detection time length, a pre-training task is used for carrying out self-supervision training on the first pre-training representation model and the first prediction model, and carrying out supervision training on the battery state prediction model according to the second battery characteristic information of the battery pack and SOH label information corresponding to the second battery characteristic information to obtain the SOH prediction model of the battery pack. The first battery characteristic information and the second battery characteristic information used in the process of training the SOH prediction model of the battery pack can be directly obtained from data reported by the EV, so that the training cost is reduced, and the EV can predict the SOH information of the battery pack according to real vehicle data.

An exemplary training method of the battery state prediction model is introduced below by taking a target battery as a battery pack, a first pre-training representation model comprising an SOC encoder and an SOC sequence encoder, a first prediction model as a classifier or a regressor, and a battery state prediction model as a fault prediction model.

Fig. 6 is a flowchart illustrating a method for training a fault prediction model of a battery pack according to an embodiment of the present application. As shown in fig. 6, the method includes at least S601 to S608.

S601, acquiring all first battery characteristic information of the battery pack in a first detection time length.

S602, dividing the first detection time into L time segments according to the SOC value of the battery pack, wherein the SOC value of the first battery characteristic information in each time segment is the same, and L is a positive integer.

And S603, inputting the first battery characteristic information in each time segment of the L time segments into an SOC encoder to obtain a representation vector corresponding to the time segment.

S604, sequentially inputting the representation vector corresponding to each time slice into the SOC sequence encoder according to the time sequence to obtain the representation vector corresponding to the first detection duration.

S605, inputting the expression vector corresponding to the first detection duration into a regressor or a classifier to obtain a pre-training output result.

S606, self-supervision training is carried out on the first pre-training representation model and the first prediction model by using a pre-training task, the first pre-training representation model comprises an SOC encoder and an SOC sequence encoder, and the first prediction model comprises a regressor or a classifier.

It should be noted that S601 to S606 may refer to S501 to S506, which is not described herein again.

S607, the second battery characteristic information of the battery pack and the fault label information corresponding to the second battery characteristic information are obtained.

And S608, performing supervision training on a battery state prediction model according to second battery characteristic information of the battery pack and fault label information corresponding to the second battery characteristic information, wherein the battery state prediction model comprises a characteristic vector extraction model and a second prediction model.

In some possible implementation manners, parameters in a feature vector extraction model in the battery state prediction model are initialized by using model parameters of a first pre-training representation model obtained after self-supervision training, and then the feature vector extraction model and the second prediction model are further trained by using second battery feature information and fault label information corresponding to the second battery feature information to obtain the battery state prediction model.

In other possible implementation manners, the first pre-training representation model obtained after the self-supervision training is used as a feature vector extraction model in the battery state prediction model, the second battery feature information is input into the feature vector extraction model to obtain a representation vector of the second battery feature information, and the second prediction model is trained by using the second battery feature information, the representation vector of the second battery feature information and the fault label information corresponding to the second battery feature information to obtain the battery state prediction model.

The battery state prediction model obtained according to the training method shown in fig. 6 may be used to predict the failure information of the battery pack.

In the technical scheme provided by the application, all first battery characteristic information of a battery pack in a first detection duration is divided into L time segments according to an SOC value of the battery pack, the battery characteristic information in each time segment of the L time segments is input into an SOC encoder in a first pre-training representation model to obtain a representation vector corresponding to each time segment, the representation vector corresponding to each time segment of the L time segments is sequentially input into an SOC sequence encoder in the first pre-training representation model according to a time sequence to obtain a representation vector corresponding to the first detection duration, the representation vector corresponding to the first detection duration is input into a first prediction model to obtain a prediction output result of the battery pack in the first detection duration, a pre-training task is used for carrying out self-supervision training on the first pre-training representation model and the first prediction model, and carrying out supervision training on the battery state prediction model according to the second battery characteristic information of the battery pack and the fault label information corresponding to the second battery characteristic information to obtain the fault prediction model of the battery pack. The first battery characteristic information and the second battery characteristic information used in the process of training the fault prediction model of the battery pack can be directly obtained from data reported by the EV, so that the training cost is reduced, and the EV can predict the fault information of the battery pack according to real vehicle data.

An exemplary training method of the battery state prediction model is described below by taking a target battery as a single battery cell, a first pre-training representation model comprising a variational encoder and a first prediction model as a variational decoder, and a battery state prediction model being an SOH prediction model.

Fig. 7 is a flowchart illustrating a method for training a SOH prediction model of a cell in accordance with an embodiment of the present disclosure, and as shown in fig. 7, the method at least includes S701 to S708.

S701, acquiring all first battery characteristic information of the single battery core within a first detection duration.

Optionally, the first battery characteristic information of the single battery cell in the first detection duration may include voltage information of the single battery cell, temperature information of the single battery cell, current information of the single battery cell, SOC information of the single battery cell, insulation resistance information of the single battery cell, or EIS information of the single battery cell.

The first battery characteristic information of the electric core may be obtained based on data reported by an original BMS of the EV.

S702, dividing the first detection duration into Q time window frames, wherein all the first battery characteristic information in each time window frame forms a corresponding expression vector of the time window frame, and Q is a positive integer.

In some possible implementations, the number of the first battery characteristic information in each of the Q time window frames may be preset, and the quantity of the first battery characteristic information in any one of the Q time window frames is equal to the quantity of the first battery characteristic information in any other one of the Q time window frames, the first battery characteristic information in the first detection duration is divided according to the quantity of the first battery characteristic information of each preset sub-time window frame, if the quantity of the first battery characteristic information in any one of the Q time window frames does not reach the quantity of the first battery characteristic information in each preset time window frame, and recording the missing first battery characteristic information as 0, wherein each time window frame comprises at least one first battery characteristic information.

The time corresponding to any first battery characteristic information in the qth time window frame in the Q time window frames in the first detection duration is earlier than the time corresponding to any first battery characteristic information in the qth +1 time window frame in the Q time window frames in the first detection duration, and Q is a positive integer smaller than Q.

S703, calculate the average value of Q representing vectors corresponding to Q time window frames.

Each of the Q time window frames corresponds to a representation vector, and the representation vector is used to represent the individual cell characteristic information of the corresponding time window frame.

S704, inputting the calculated average value of the Q expression vectors into a variation encoder to obtain an expression vector corresponding to the first detection time length.

And the variation encoder obtains a representation vector corresponding to the first detection time length according to the average value of the Q representation vectors, and the representation vector corresponding to the first detection time length is used for representing the monomer electric core characteristic information of the monomer electric core in the first detection time length.

And S705, inputting the expression vector corresponding to the first detection duration into a variation decoder to obtain a prediction output result of the single battery cell within the first detection duration.

S706, self-supervision training is carried out on the first pre-training representation model and the first prediction model by using a pre-training task, the first pre-training representation model comprises a variation encoder, and the first prediction model comprises a variation decoder.

This step may refer to S506, which is not described herein.

And S707, acquiring second battery characteristic information of the single battery cell and SOH label information corresponding to the second battery characteristic information.

Optionally, the second battery characteristic information of the single battery cell within the first detection duration may include voltage information of the single battery cell, temperature information of the single battery cell, current information of the single battery cell, SOC information of the single battery cell, insulation resistance information of the single battery cell, or EIS information of the single battery cell.

And S708, performing supervision training on a battery state prediction model according to second battery characteristic information of the single battery cell and SOH label information corresponding to the second battery characteristic information, wherein the battery state prediction model comprises a characteristic vector extraction model and a second prediction model.

The battery state prediction model obtained according to the training method shown in fig. 7 may be used to predict SOH information of the individual battery cells.

In the technical scheme provided by the application, all first battery characteristic information of a single battery cell in a first detection time length is divided into Q time window frames, all first battery characteristic information in each time window frame forms a representation vector corresponding to the time window frame, an average value of Q representation vectors corresponding to the Q time window frames is calculated, the average value is input into a first pre-training representation model to obtain a representation vector corresponding to the first detection time length, the representation vector corresponding to the first detection time length is input into a first prediction model to obtain a prediction output result of the single battery cell in the first detection time length, a pre-training task is used for carrying out self-supervision training on the first pre-training representation model and the first prediction model, and supervision training is carried out on a battery state prediction model according to second battery characteristic information of the single battery cell and SOH label information corresponding to the second battery characteristic information, and obtaining the SOH prediction model of the single battery cell. The first battery characteristic information and the second battery characteristic information used in the process of training the SOH prediction model of the single battery cell can be directly obtained from data reported by the EV, so that the training cost is reduced, and the EV can predict the SOH information of the single battery cell according to real vehicle data.

An exemplary training method of the battery state prediction model is described below by taking a target battery as a single battery cell, a first pre-training representation model comprising a variational encoder, a first prediction model being a variational decoder, and a battery state prediction model being a fault prediction model.

Fig. 8 is a schematic flowchart of a method for training a fault prediction model of a cell in accordance with an embodiment of the present application, and as shown in fig. 8, the method at least includes S801 to S808.

S801, acquiring all first battery characteristic information of the single battery core within a first detection duration.

S802, dividing the first detection duration into W time window frames, wherein all the first battery characteristic information in each time window frame forms a corresponding expression vector of the time window frame.

S803, an average of W representative vectors corresponding to the W time window frames is calculated.

S804, inputting the calculated average value of the W expression vectors into a variation encoder to obtain the expression vector corresponding to the first detection time length.

And S805, inputting the expression vector corresponding to the first detection time length into a variational decoder to obtain predicted target battery characteristic information.

S806, self-supervised training is carried out on the first pre-training representation model and the first prediction model by using a pre-training task, the first pre-training representation model comprises a variational encoder, and the first prediction model comprises a variational decoder.

It should be noted that S801 to S806 may refer to S701 to S706, which is not described herein again.

And S807, acquiring second battery characteristic information of the single battery core and fault label information corresponding to the second battery characteristic information.

Optionally, the second battery characteristic information of the single battery cell within the duration of the first detection may include voltage information of the single battery cell, temperature information of the single battery cell, current information of the single battery cell, SOC information of the single battery cell, insulation resistance information of the single battery cell, or EIS information of the single battery cell, and the like.

And S808, performing supervision training on a battery state prediction model according to second battery characteristic information of the single battery cell and fault label information corresponding to the second battery characteristic information, wherein the battery state prediction model comprises a characteristic vector extraction model and a second prediction model.

The battery state prediction model obtained according to the training method shown in fig. 8 may be used to predict the fault information of the individual battery cells.

In the technical scheme provided by the application, all first battery characteristic information of a single battery cell in a first detection time length is divided into W time window frames, all first battery characteristic information in each time window frame forms a representation vector corresponding to the time window frame, an average value of W representation vectors corresponding to the W time window frames is calculated, the average value is input into a first pre-training representation model to obtain a representation vector corresponding to the first detection time length, the representation vector corresponding to the first detection time length is input into a first prediction model to obtain a prediction output result of the single battery cell in the first detection time length, a pre-training task is used for carrying out self-supervision training on the first pre-training representation model and the first prediction model, and supervision training is carried out on a battery state prediction model according to second battery characteristic information of the single battery cell and fault label information corresponding to the second battery characteristic information, and obtaining a fault prediction model of the single battery cell. The first battery characteristic information and the second battery characteristic information used in the process of training the fault prediction model of the single battery cell can be directly obtained from data reported by the EV, so that the training cost is reduced, and the EV can predict the fault information of the single battery cell according to real vehicle data.

Fig. 9 is a flowchart illustrating a method for predicting a battery state according to an embodiment of the present application, and as shown in fig. 9, the method at least includes S901 to S902.

And S901, acquiring the characteristic information of the battery to be predicted.

The battery to be predicted may be a battery pack including a plurality of battery cells, or may be a battery cell.

As an example, when the battery to be predicted is a battery pack, the characteristic information of the battery to be predicted may include current information of the battery pack, voltage information of the battery pack, temperature information of the battery pack, voltage information of each cell in the battery pack, SOC information of the battery pack, and the like.

As another example, when the battery to be predicted is a single battery cell, the battery characteristic information to be predicted may include voltage information of the single battery cell, temperature information of the single battery cell, current information of the single battery cell, SOC information of the single battery cell, insulation resistance information of the single battery cell, or EIS information of the single battery cell.

In some implementation manners, the battery characteristic information to be predicted of the battery to be predicted may be directly obtained from data reported by an original BMS of an EV to which the battery to be predicted belongs.

And S902, determining the battery state information of the battery to be predicted based on the characteristic information of the battery to be predicted by using a battery state prediction model.

The battery state information of the battery to be predicted can be obtained by inputting the characteristic information of the battery to be predicted into the battery state prediction model, and the state information of the battery to be predicted can comprise SOH information and/or fault information and/or residual life information of the battery to be predicted and the like.

It is understood that the battery state prediction model may be trained according to the training methods shown in fig. 4 to 8.

For example, when the battery to be predicted is a battery pack, the SOH information of the battery pack may be obtained by predicting using a battery state prediction model obtained by the training method shown in fig. 5; the failure information of the battery pack can be obtained by predicting the battery state prediction model obtained by the training method shown in fig. 6.

For example, when the battery to be predicted is a single battery cell, the SOH information of the single battery cell may be obtained by predicting using the battery state prediction model obtained by the training method shown in fig. 7; the battery state prediction model obtained by using the training method shown in fig. 8 is used for prediction, so that the fault information of the single battery cell can be obtained.

In the technical scheme provided by the application, the characteristic information of the battery to be predicted is input into the battery state prediction model obtained by training according to the training method described in any one of fig. 4 to 8, so as to obtain the battery state information of the battery to be predicted. The battery characteristic information to be predicted of the battery to be predicted can be directly obtained from data reported by the EV, so that the EV can predict the battery state according to real vehicle data.

Fig. 10 is a schematic configuration diagram of a training device of a battery state prediction model according to an embodiment of the present application. As shown in fig. 10, the apparatus 1000 may include an acquisition module 1001 and a training module 1002.

Any module in the acquisition module and the training module in the embodiments of the present application may be wholly or partially implemented by software and/or hardware. The part realized by software can be run on the processor to realize corresponding functions, and the part realized by hardware can be a constituent part of the processor.

In one implementation, the apparatus 1000 may be used to implement the method illustrated in FIG. 4 described above. For example, the obtaining module 1001 is configured to implement S401 and S403, and the training module 1002 is configured to implement S402 and S404.

In another implementation, the apparatus 1000 may be used to implement the method illustrated in FIG. 5 described above. For example, the obtaining module 1001 is configured to implement S501 and S507, and the training module 1002 is configured to implement S506 and S508.

In yet another implementation, the apparatus 1000 may be used to implement the method illustrated in FIG. 6 described above. For example, the obtaining module 1001 is configured to implement S601 and S607, and the training module 1002 is configured to implement S606 and S608.

In yet another implementation, the apparatus 1000 may be used to implement the method illustrated in FIG. 7 described above. For example, the obtaining module 1001 is configured to implement S701 and S707, and the training module 1002 is configured to implement S706 and S708.

In yet another implementation, the apparatus 1000 may be used to implement the method illustrated in FIG. 8 described above. For example, the obtaining module 1001 is configured to implement S801 and S807, and the training module 1002 is configured to implement S806 and S808.

Fig. 11 is a schematic configuration diagram of a battery state prediction apparatus according to an embodiment of the present application. As shown in fig. 11, the apparatus 1100 may include an acquisition module 1101 and a processing module 1102.

Any module in the acquisition module and the processing module in the embodiments of the present application may be wholly or partially implemented by software and/or hardware. The part realized by software can be run on the processor to realize corresponding functions, and the part realized by hardware can be a constituent part of the processor.

In one implementation, the apparatus 1100 may be used to implement the method illustrated in FIG. 9 described above. For example, the acquiring module 1101 is used to implement S901, and the processing module 1102 is used to implement S902.

Fig. 12 is a schematic structural diagram of an apparatus according to an embodiment of the present application. The apparatus 1200 shown in fig. 12 may be used to perform the method described in any of the previous embodiments.

As shown in fig. 12, the apparatus 1200 of the present embodiment includes: a memory 1201, a processor 1202, a communication interface 1203, and a bus 1204. The memory 1201, the processor 1202, and the communication interface 1203 are communicatively connected to each other through a bus 1204.

The memory 1201 may be a Read Only Memory (ROM), a static memory device, a dynamic memory device, or a Random Access Memory (RAM). The memory 1201 may store programs and the processor 1202 may be configured to perform the steps of the methods shown in fig. 4-9 when the programs stored in the memory 1201 are executed by the processor 1202.

The processor 1202 may be a general-purpose Central Processing Unit (CPU), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits, configured to execute related programs to implement the method for training the battery state prediction model and the method for predicting the battery state according to the embodiment of the present invention.

The processor 1202 may also be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the method of the embodiments of the present application may be implemented by integrated logic circuits of hardware or instructions in the form of software in the processor 1202.

The processor 1202 may also be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, or discrete hardware components. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The steps of the method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in the memory 1201, and the processor 1202 reads the information in the memory 1201, and completes the functions required to be performed by the methods in the embodiments of the present application in combination with hardware of the processor 1202, for example, may perform the steps/functions of the embodiments shown in fig. 4 to 9.

The communication interface 1203 may use transceiver means such as, but not limited to, a transceiver to enable communication between the apparatus 1200 and other devices or communication networks.

The bus 1204 may include pathways to transfer information between various components of the apparatus 1200 (e.g., memory 1201, processor 1202, communication interface 1203).

It should be understood that the apparatus 1200 shown in the embodiment of the present application may be an electronic device, or may also be a chip configured in the electronic device.

It should be understood that the processor in the embodiments of the present application may be a Central Processing Unit (CPU), and the processor may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

It will also be appreciated that the memory in the embodiments of the subject application may be either volatile memory or nonvolatile memory, or may include both volatile and nonvolatile memory. The non-volatile memory may be a read-only memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an electrically Erasable EPROM (EEPROM), or a flash memory. Volatile memory can be Random Access Memory (RAM), which acts as external cache memory. By way of example, but not limitation, many forms of Random Access Memory (RAM) are available, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct bus RAM (DR RAM).

The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, the above-described embodiments may 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 or computer programs. The procedures or functions according to the embodiments of the present application are wholly or partially generated when the computer instructions or the computer program are loaded or executed on a computer. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored on a computer readable storage medium or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, data center, etc., that contains one or more collections of available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium may be a solid state disk.

It should be understood that the term "and/or" herein is only one kind of association relationship describing the association object, and means that there may be three kinds of relationships, for example, a and/or B, and may mean: a exists singly, A and B exist simultaneously, and B exists singly, wherein A and B can be singular or plural. In addition, the "/" in this document generally indicates that the former and latter associated objects are in an "or" relationship, but may also indicate an "and/or" relationship, which may be understood with particular reference to the former and latter text.

In the present application, "at least one" means one or more, "a plurality" means two or more. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of the singular or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, a-b, a-c, b-c, or a-b-c, wherein a, b, c may be single or multiple.

It should be understood that, in the various embodiments of the present application, the sequence numbers of the above-mentioned processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

It can be clearly understood by those skilled in the art that, for convenience and simplicity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: u disk, removable hard disk, read only memory, random access memory, magnetic or optical disk, etc. for storing program codes.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A training method of a battery state prediction model is characterized by comprising the following steps:

acquiring first battery characteristic information of a target battery in a first time period, wherein the battery characteristic information of the target battery comprises one or more of the following information: the current information of the target battery, the voltage information of the target battery, the temperature information of the target battery and the battery power index SOC of the target battery are obtained;

performing self-supervision training on a first model according to the first battery characteristic information, wherein the first model comprises a first pre-training representation model and a first prediction model, the input of the first prediction model comprises the output of the first pre-training representation model, the first pre-training representation model is used for determining a representation vector of the input battery characteristic information, and the first prediction model is used for determining target battery characteristic information corresponding to the input representation vector;

acquiring second battery characteristic information and first label information of the target battery, wherein the first label information is used for indicating battery state information corresponding to the second battery characteristic information;

and performing supervision training on the battery state prediction model according to the second battery characteristic information and the first label information, wherein the battery state prediction model comprises a characteristic vector extraction model and a second prediction model, the input of the second prediction model comprises the output of the characteristic vector extraction model, and the initial parameters of the characteristic vector extraction model comprise the parameters of the first pre-training expression model obtained after the self-supervision training is performed on the first model.

2. A training method according to claim 1, wherein the structure of the second predictive model is the same as that of the first predictive model, and the initial parameters of the second predictive model before the supervised training are comprised of the parameters of the first predictive model after the unsupervised training of the first model.

3. Training method according to claim 1 or 2, wherein the first battery characteristic information comprises N battery characteristic information of the target battery at N times within the first time period, N being a positive integer, wherein the first pre-trained representation model comprises a first encoder and a second encoder, wherein an input of the first encoder comprises the battery characteristic information and an input of the second encoder comprises an output of the first encoder.

4. The training method of claim 3, wherein the self-supervised training of the first model based on the first battery characteristic information comprises:

dividing the N pieces of battery characteristic information into N subsets, wherein each subset of the N subsets comprises at least one piece of battery characteristic information, the time corresponding to any battery characteristic information in the ith subset of the N subsets in the first time is earlier than the time corresponding to any battery characteristic information in the (i + 1) th subset of the N subsets in the first time, N is a positive integer and N is less than or equal to N, i is a positive integer and i is less than N;

inputting a battery characteristic information sequence obtained by arranging all the battery characteristic information in each subset of the n subsets according to the corresponding time sequence into the first encoder to obtain a representation vector corresponding to each subset;

inputting a battery characteristic information sequence obtained after arranging the n expression vectors corresponding to the n subsets according to a corresponding time sequence or a sequence from low SOC value to high SOC value into the second encoder to obtain a first expression vector of the first battery characteristic information;

inputting the first expression vector into the first prediction model to obtain target battery characteristic information corresponding to the first expression vector;

and adjusting parameters of the first model according to the target battery characteristic information and the first battery characteristic information.

5. The training method according to claim 4, wherein the dividing the N battery characteristic information into N subsets comprises:

and dividing the battery characteristic values into n subsets according to the SOC values, wherein the SOC values corresponding to all battery characteristic information contained in each of the n subsets are the same.

6. Training method according to any of the claims 3 to 5, wherein the target battery is a battery pack, and the first and/or second battery characteristic information further comprises: and voltage information of each single battery cell in the battery pack.

7. The training method of claim 1, wherein the first battery characteristic information comprises M battery characteristic information of the target battery at M times within a first period, M being a positive integer, wherein the first pre-trained representation model comprises a variational encoder, wherein the first prediction model comprises a variational decoder, and wherein the second prediction model comprises a regression model or a classification model.

8. The training method of claim 7, wherein the self-supervised training of the first model based on the first battery characteristic information comprises:

dividing the M pieces of battery characteristic information into M subsets, wherein each subset in the M subsets comprises at least one piece of battery characteristic information, the time corresponding to any battery characteristic information in the jth subset in the M subsets in the first time is earlier than the time corresponding to any battery characteristic information in the (j + 1) th subset in the M subsets in the first time, M is a positive integer and M is less than or equal to M, j is a positive integer and j is less than M;

determining a representation vector corresponding to each subset according to all battery characteristic information in each subset of the m subsets;

calculating the average value of m representing vectors corresponding to the m subsets;

inputting the average value into the first pre-training representation model to obtain a second representation vector of the first battery characteristic information;

inputting the second expression vector into the first prediction model to obtain target battery characteristic information corresponding to the second expression vector;

9. The training method according to claim 8, wherein the number of the battery characteristic information in any one of the m subsets is equal to the number of the battery characteristic information in any other one of the m subsets.

10. The training method according to any one of claims 7 to 9, wherein the target battery is a single battery cell, and the first battery characteristic information and/or the second battery characteristic information further include: and the insulation resistance information of the single battery cell and/or the electrochemical alternating current impedance spectrum EIS of the single battery cell.

11. Training method according to any of claims 1 to 10, wherein the target battery characteristic information comprises: the SOC sequence of the target battery in the first time period, the voltage sequence of the target battery in the first time period, the charging time length sequence of each cell in the target battery in the first time period, the charging mode of the target battery, or the SOC value of the target battery at a random mask in the first time, where the charging mode includes a fast charging mode or a slow charging mode.

12. The method according to any one of claims 1 to 11, characterized in that the battery state information comprises battery health indicator SOH information or fault information or remaining life information.

13. A method for predicting a state of a battery, comprising:

acquiring characteristic information of a battery to be predicted of the battery to be predicted;

determining battery state information of the battery to be predicted based on the battery characteristic information to be predicted by using a battery state prediction model, wherein the battery state prediction model is obtained by training by using the training method of any one of claims 1 to 12.

14. A training device for a battery state prediction model, characterized by comprising functional modules for implementing the method of any one of claims 1 to 12.

15. A battery state prediction apparatus comprising functional blocks for implementing the method of claim 13.

16. A training apparatus for a battery state prediction model, comprising: a memory and a processor;

the memory is to store program instructions;

the processor is configured to invoke program instructions in the memory to perform the method of any of claims 1 to 12.

17. An apparatus for predicting a state of a battery, comprising: a memory and a processor;

the memory is to store program instructions;

the processor is configured to call program instructions in the memory to perform the method of claim 13.

18. A chip comprising at least one processor and a communication interface, the communication interface and the at least one processor being interconnected by a line, the at least one processor being configured to execute a computer program or instructions to perform the method of any one of claims 1 to 12 or the method of claim 13.

19. A computer-readable medium, characterized in that the computer-readable medium stores program code for computer execution, the program code comprising instructions for performing the method of any one of claims 1 to 12 or the method of claim 13.

20. A computer program product comprising instructions which, when executed, cause a computer to perform the method of any one of claims 1 to 12 or the method of claim 13.