Method, device, and device for predicting power battery capacity

Abstract

A method for predicting power battery capacity comprises: a step of acquiring sample data of a power battery; a step of dividing the sample data into multiple categories using a clustering algorithm, wherein each category has a corresponding aging model and a feature identifier; a step of acquiring battery state parameters of a power battery to be tested; a step of determining an aging model to be used by the battery state parameters from a plurality of aging models; a step of inputting the battery state parameters into the aging model used and a step of obtaining a corresponding battery capacity, wherein the aging model may be obtained by the following steps: a step of determining a fitting relationship between aging models, and a step of determining parameters of the fitting relationship by sample data of a corresponding type of aging model. A power battery capacity prediction device and device, and a corresponding storage medium are also disclosed.

Description

Technology Field

[0001] Cross-reference regarding related applications

[0002] The present disclosure claims priority and advantage to Chinese Patent Application No. 202110711103.9, filed on June 25, 2021, which is incorporated herein by reference as is and is titled “Method, Apparatus, and Device for Predicting Capacity of Power Battery”.

[0003] field

[0004] The present disclosure relates to the technical field of battery management, and more specifically, to a method for predicting the capacity of a power battery, an apparatus for predicting the capacity of a power battery, a device for predicting the capacity of a power battery, and a corresponding storage medium. Background Technology

[0005] Power batteries are a critical component of electric vehicles. Lifespan is a key performance indicator for power batteries. Accurately predicting lifespan not only helps understand battery degradation, provides users with precise information on vehicle operating status, and serves as a basis for cost estimation in vehicle production and manufacturing, but also assists in preventing defects and accidents, thereby ensuring the safety of users' lives and property.

[0006] The lifespan of power batteries is typically estimated by experimental and model methods.

[0007] In experimental methods, the current of the power battery is not constant during the actual operation of the vehicle, leading to inaccurate prediction results. Furthermore, while standard new European driving cycle (NEDC) conditions may be adopted to simulate actual operating conditions for discharge tests, the test cycle is too long.

[0008] Model methods primarily adopt mechanistic and statistical models. Mechanistic models include electrochemical analysis methods and impedance methods. Statistical models mainly refer to life prediction models designed based on neural networks, such as life prediction methods based on Long Short-Term Memory (LSTM) neural networks and transfer learning, as well as deep learning methods for predicting the lifespan of lithium batteries. Actual vehicle driving conditions are complex, mechanistic model parameters are difficult to obtain, and accurately predicting battery lifespan is challenging. Prior art literature

[65535] US 2020 / 284846 A1 (Published: Sep. 10, 2020)CN 112505569 A (Published: Mar. 16, 2021)US 2016 / 239592 A1 (Published: Aug. 18, 2016)CN 103954913 A (Published: Jul. 30, 2014)US 2013 / 090900 Al (Published: Apr. 11, 2013)CN 111537884 A (Published: Aug. 14, 2020) The problem to be solved

[0009] In this regard, the present disclosure aims to propose a method, apparatus, and device for predicting the capacity of a power battery to at least partially solve the problems of the prior art, such as the long test cycle of experimental methods and the difficulty in obtaining parameters and high model complexity of model methods. means of solving the problem

[0010] To achieve the above objective, the present disclosure provides a method for predicting the capacity of a power battery. The prediction method comprises the following: Sample data of a power battery is obtained. The sample data is divided into several categories using a clustering algorithm, and each category has a corresponding aging model and a feature identifier. The aging model is obtained by the following steps: The fitting relationship of the aging model is determined. The parameters of the fitting relationship are determined according to the sample data of the aging model of the corresponding type. The feature identifier is used to identify the features of the sample data of the corresponding category. Battery state parameters of the power battery to be tested are obtained. The aging model adopted by the battery state parameters is determined from a plurality of aging models. The battery state parameters are input into the adopted aging model to obtain the corresponding battery capacity.

[0011] According to an embodiment of the present disclosure, sample data of a power battery includes a plurality of sets of historical data for the same model of power battery under actual vehicle driving conditions.

[0012] According to an embodiment of the present disclosure, the step of dividing sample data into multiple categories using a clustering algorithm comprises the following: the clustering algorithm is pre-set, and the clustering parameters of the clustering algorithm are determined. The sample data is divided into core points and boundary points according to the clustering parameters. Categories are organized according to the core points, and the sample data is divided into multiple categories.

[0013] According to an embodiment of the present disclosure, the clustering algorithm is the DBSCAN algorithm; and the clustering parameters include a radius of neighbors and a neighbor count threshold.

[0014] According to an embodiment of the present disclosure, the feature identifier is a clustering center; the feature identifier is a clustering center. The step of determining an aging model adopted by a battery state parameter from a plurality of aging models comprises: the distance between the battery state parameter and the clustering center corresponding to each category is calculated. The closest aging model is selected as the aging model adopted by the battery state parameter.

[0015] According to an embodiment of the present disclosure, the fitting relationship includes polynomial fitting, neural network fitting, or regression tree fitting.

[0016] According to an embodiment of the present disclosure, battery state parameters include at least two of current, voltage, temperature, charge state, storage time, discharge depth, and Coulomb efficiency.

[0017] In a second aspect of the present disclosure, an apparatus for predicting the capacity of a power battery is also provided, comprising: an input unit configured to obtain battery state parameters; a matcher configured to match the battery state parameters to determine an aging model adopted by the battery state parameters from a plurality of aging models, wherein the plurality of aging models correspond one-to-one with several categories of sample data partitioned by a clustering algorithm, and each aging model includes a mapping relationship between the battery state parameters and the battery capacity; and a calculator configured to input the battery state parameters into the adopted aging model to obtain the corresponding battery capacity.

[0018] In a third aspect of the present disclosure, a device for predicting the capacity of a power battery is also provided, comprising at least one processor and a memory connected to at least one processor. The memory stores instructions executable by at least one processor, and the at least one processor implements the method for predicting the capacity of the power battery described above by executing instructions stored in the memory.

[0019] In a fourth aspect of the present disclosure, a computer-readable storage medium is also provided and stores a computer program. When the program is executed by a processor, it implements the aforementioned method for predicting the capacity of a power battery.

[0020] Compared to the prior art, the method, apparatus, and device for predicting the capacity of a power battery according to an embodiment of the present disclosure have the following beneficial effects:

[0021] Through the embodiment provided by the present disclosure, multiple aging types can be distinguished, and thus, corresponding aging models are set according to different aging types, thereby improving the precision of the aging models and predicting the capacity of the power battery more accurately. As sample data becomes richer, there exists an increasing number of data for different aging types and different aging types with wider coverage, and the clustering algorithm can distinguish different categories more comprehensively, and the distinction effect is more intuitive and reliable.

[0022] Other features and advantages of the present disclosure will be described in detail in the following detailed description. Brief explanation of the drawing

[0023] The accompanying drawings, which constitute part of the present disclosure, are used to provide further understanding of the present disclosure. Exemplary embodiments of the present disclosure and descriptions thereof are used to explain the present disclosure and do not constitute an inappropriate limitation to the present disclosure. In the accompanying drawings: FIG. 1 is a schematic flowchart of a method for predicting the capacity of a power battery according to an embodiment of the present disclosure. FIG. 2 is a flowchart of a clustering algorithm in a method for predicting the capacity of a power battery according to an embodiment of the present disclosure. FIG. 3 is a schematic diagram of the calculation of clustering center distance in a method for predicting the capacity of a power battery according to an embodiment of the present disclosure. FIG. 4 is a schematic flowchart of an embodiment of a method for predicting the capacity of a power battery according to an embodiment of the present disclosure. FIG. 5 is a schematic diagram of a device for predicting the capacity of a power battery according to an embodiment of the present disclosure. Specific details for implementing the invention

[0024] It should be noted that embodiments of the present disclosure and features of the embodiments may be combined with one another where they do not conflict.

[0025] The present disclosure is described in detail with reference to the accompanying drawings and by combining the following embodiments.

[0026] FIG. 1 is a schematic flowchart of a method for predicting the capacity of a power battery according to an embodiment of the present disclosure, as illustrated in FIG. 1. A method for predicting the capacity of a power battery is provided. The prediction method comprises the following:

[0027] S01: Sample data of the power battery is obtained.

[0028] Sample data can be obtained from a power battery under actual vehicle driving conditions, and the sample data may include current (I), voltage (V), temperature (T), state of charge (SOC), storage time (t), depth of discharge (DOD), and Coulomb efficiency (μ), or a combination of parameters selected from these.

[0029] S02: Sample data is divided into multiple categories using a clustering algorithm, and a corresponding aging model and feature identifier are determined for each category. The aging model is obtained through the following steps: The fitting relationship of the aging model is determined. The parameters of the fitting relationship are determined based on the sample data of the aging model of the corresponding type. The feature identifier is used to identify the features of the sample data of the corresponding category.

[0030] Sample data is classified, and each category of the sample data possesses a specific similarity or implicit correlation. By classifying the sample data through a clustering algorithm, classification results can be obtained rapidly and are desirable. The clustering algorithm may be selected from existing clustering algorithms depending on actual needs. An aging model is required to be determined for each category. The aging model is a mathematical model, and first, it is necessary to determine the fitting relationship within the mathematical model, that is, to select an appropriate fitting function. Subsequently, the fitting relationship is trained or calibrated by the sample data within the category to determine the parameters of the fitting relationship, and thus, an aging model for calculating battery capacity based on the input parameters is obtained.

[0031] S03: Battery state parameters of the power battery to be tested are obtained.

[0032] The battery state parameters obtained here serve as input parameters for predicting capacity, and include parameters that are identical to or have a subset relationship with the parameters of the sample data in step S01.

[0033] S04: The aging model adopted by the battery state parameter is determined from a plurality of aging models.

[0034] It is necessary to first determine the aging model adopted for a specific battery state parameter, and based on the same battery state parameter, different battery capacities may be obtained according to different aging models. In this embodiment, the adopted aging model is determined according to a feature identifier. By determining an appropriate aging model, a more accurate battery capacity can be calculated.

[0035] S05: Battery state parameters are input into the adopted aging model to obtain the corresponding battery capacity. The aging model adopted at this step is the aging model determined in step S04, and is configured to obtain the corresponding battery capacity according to the battery state parameters, that is, the corresponding battery capacity can be obtained by inputting the battery state parameters.

[0036] Through the above embodiment, multiple aging types can be distinguished, and thus, corresponding aging models are set according to different aging types, thereby improving the precision of the aging model and predicting the capacity of the power battery more accurately. As sample data becomes richer, there exists an increasing number of data for different aging types and different aging types with wider coverage, and the clustering algorithm can distinguish different categories more comprehensively, and the distinction effect is more intuitive and reliable.

[0037] In an embodiment provided by the present disclosure, sample data of a power battery includes multiple sets of historical data for the same model of the power battery under actual vehicle driving conditions. Historical data under actual vehicle driving conditions is adopted as a sample, which better reflects the actual scenario and facilitates the acquisition of a large number of samples. Through a large number of sample data reflecting the actual state of multiple power batteries, the limitations of experimental and model methods are overcome, which enables more accurate clustering and thus more accurate prediction of battery capacity.

[0038] FIG. 2 is a flowchart of a clustering algorithm in a method for predicting the capacity of a power battery according to an embodiment of the present disclosure, as illustrated in FIG. 2. In this embodiment, the step of dividing sample data into several categories using a clustering algorithm comprises the following: A clustering algorithm is pre-set, and clustering parameters of the clustering algorithm are determined. Sample data is divided into core points and boundary points according to the clustering parameters. Categories are configured according to the core points, and sample data is divided into several categories. Additionally, the clustering algorithm is a DBSCAN algorithm; and the clustering parameters include a radius of neighbors (Eps) and a neighbor count threshold (Minpts). The specific process is as follows: In an embodiment, a data set is first scanned, an unvisited point (p) is selected, and a neighbor set (Np) is generated. If the count within Eps(p) is greater than Minpts, p is determined as a core point, and a new cluster (C) is generated. Subsequently, an unclassified point (q) within Np is selected. If q is not visited, a neighbor set (Nq) is created. If the count in Eps(q) is greater than Minpts, q is determined to be a core point, and Np is updated as Np = Np + Nq. Point (q) is added to cluster (C), and if q is neither a core point nor assigned to any category, q is determined to be a boundary point to be added to cluster (C). The process continues until Np no longer contains unclassified points. After this, if there are still unvisited points in the dataset (D), the second step is repeated, and the unvisited points are selected. Through the above method, the sample data is divided into the aforementioned categories.

[0039] FIG. 3 is a schematic diagram of the calculation of clustering center distances in a method for predicting the capacity of a power battery according to an embodiment of the present disclosure, as illustrated in FIG. 3. In this embodiment, the feature identifier is a clustering center. The step of determining an aging model adopted by a battery state parameter from a plurality of aging models comprises the following: the distance between the battery state parameter and the clustering center corresponding to each category is calculated. The closest aging model is selected as the aging model adopted by the battery state parameter. The figure illustrates only the situation of four clustering centers (C1 to C4), and the number of clustering centers is not limited to the number of types. Conventional clustering calculations in clustering analysis include Euclidean distance, Manhattan distance, Chebyshev distance, etc., which are mainly used to measure similarity. By the calculation according to the above method, similarity (distance) between two objects can be obtained. In the actual calculation, an effective selection is made based on the attribute characteristics of different objects. Since the calculation of distances between objects directly affects the efficiency of the algorithm, it is very important in the clustering algorithm process; therefore, careful consideration is required when making actual selections.

[0040] In the embodiments provided by the present disclosure, the fitting relationship includes polynomial fitting, neural network fitting, or regression tree fitting. Polynomial fitting includes: y = p_{0}x^n + p_{1}x^{n-1} + p_{2}x^{n-2} + p_{3}x^{n-3} + ... + p_{n}. The number of terms in the polynomial may be determined as needed. Neural network fitting includes: convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), etc. By selecting an appropriate neural network structure and training the neural network structure with sample data, an aging model capable of predicting battery capacity can be obtained. Regression tree fitting includes a common binary tree. For example, the binary tree is used to recursively partition the prediction space into several subsets, and the distribution of Y within these subsets is continuous and uniform. Leaf nodes within the tree correspond to different partitioned regions, and the partitioning is determined by the partitioning rule associated with each internal node. By traversing from the root to the leaf nodes, prediction samples are assigned unique leaf nodes, and the conditional distribution of Y at these nodes is also determined. Specific setup steps for different aging models are not repeated here.

[0041] In an embodiment provided by the present disclosure, the battery state parameters include at least two of current, voltage, temperature, state of charge, storage time, depth of discharge, and Coulomb efficiency. The more input parameters there are, the more accurate the obtained battery capacity will be. In a specific scenario, a person skilled in the art selects at least two of the battery state parameters for combination based on actual conditions and measurement conditions to obtain a more accurate battery capacity.

[0042] FIG. 4 is a schematic flowchart of an embodiment of a method for predicting the capacity of a power battery according to an embodiment of the present disclosure, as illustrated in FIG. 4. In this embodiment, the method for predicting the capacity of a power battery comprises the following steps:

[0043] (1) Clustering algorithm, parameters, and related thresholds are pre-set.

[0044] (2) Sample data of the power battery is input under actual vehicle driving conditions.

[0045] (3) Sample data is divided into several categories through a clustering algorithm, and clustering centers (C1, ..., Ck) are obtained.

[0046] (4) Model parameters corresponding to the aging model are obtained by performing statistical models such as polynomial fitting, neural network fitting, and regression tree fitting on each category of the sample data.

[0047]

[0048] (5) To determine which aging model the data to be tested belongs to, the distance between the data to be tested and the clustering center is compared.

[0049] (6) Data to be tested to calculate the capacity of the power battery is input into the corresponding aging model.

[0050] FIG. 5 is a schematic diagram of an apparatus for predicting the capacity of a power battery according to an embodiment of the present disclosure, as illustrated in FIG. 5. In this embodiment, the apparatus for predicting the capacity of a power battery comprises: an input unit configured to obtain battery state parameters; a matcher configured to match the battery state parameters to determine an aging model adopted by the battery state parameters from a plurality of aging models, wherein the plurality of aging models correspond one-to-one with several categories of sample data divided by a clustering algorithm, and each aging model includes a mapping relationship between the battery state parameters and the battery capacity; and a calculator configured to input the battery state parameters to the adopted aging model to obtain the corresponding battery capacity.

[0051] Specific limitations regarding various modules (input unit, matcher, calculator) within the device for predicting the capacity of a power battery may refer to the limitations regarding the method for predicting the capacity of a power battery described above, and are not repeated herein. The various modules within the device may be implemented wholly or partially by software, hardware, or a combination thereof. The various modules may be embedded in the processor of a computer device in the form of hardware or independently, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute corresponding operations of the various modules.

[0052] In an embodiment provided by the present disclosure, a device for predicting the capacity of a power battery is also provided and comprises at least one processor and memory connected to at least one processor. The memory stores instructions executable by at least one processor, and the at least one processor implements the method for predicting the capacity of the power battery described above by executing instructions stored in the memory. The controller or processor mentioned herein has the functions of numerical calculation and logical operation and has at least a data processing capability and has a central processing unit (CPU), random access memory (RAM), read-only memory (ROM), multiple I / O ports and an interrupt system, etc. The processor includes a core, and the core calls a corresponding program unit from memory. There may be one or more cores, and the method described above is implemented by adjusting core parameters. The memory may include forms such as volatile memory, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory (Flash RAM) in a computer-readable medium.

[0053] In an embodiment provided by the present disclosure, a computer-readable storage medium stores a computer program. When executed by a processor, the computer program implements the aforementioned method for predicting the capacity of a power battery.

[0054] A person skilled in the art should understand that embodiments of the present disclosure may be provided as a method, system, or computer program product. Accordingly, the present disclosure may be used in the form of embodiments for hardware only, embodiments for software only, embodiments combining software and hardware, etc. Furthermore, the present disclosure may be used in the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0055] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products in embodiments of the present disclosure. It is understood that computer program instructions may implement each procedure and / or block of the flowchart and / or block diagram, and combinations of procedures and / or blocks of the flowchart and / or block diagram. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that a device configured to implement the functions specified in one or more procedures of the flowchart and / or one or more blocks of the block diagram is generated using instructions executed by the processor of the computer or other programmable data processing device.

[0056] These computer program instructions, which can command a computer or other programmable data processing device to operate in a specific manner, may alternatively be stored in computer-readable memory, and instructions stored in computer-readable memory create an artifact comprising an instruction unit. The instruction unit implements functions specified in one or more procedures of a flowchart and / or one or more blocks of a block diagram.

[0057] These computer program instructions may be additionally loaded into a computer or other programmable data processing device, so that a series of operations and steps are performed on the computer or other programmable device, thereby generating computer-implemented processing. Accordingly, instructions executed on the computer or other programmable device provide steps for implementing functions specified in one or more procedures of a flowchart and / or one or more blocks of a block diagram.

[0058] In a typical configuration, a computer device includes one or more central processing units (CPUs), input / output interfaces, network interfaces, and internal memory.

[0059] Memory may include forms such as volatile memory, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory (Flash RAM) within a computer-readable medium. Memory is an example of a computer-readable medium.

[0060] Computer-readable media include non-volatile and volatile media, removable media and non-removable media, which may also implement the storage of information using any method or technique. The information may be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), or other types of random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EEPROM), flash memory, or other storage technologies, compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), or other optical storage devices, cartridge tape, magnetic tape, magnetic disk storage device, or other magnetic storage device, or any other non-transmission media that may be configured to store information accessible by a computing device. According to the limitations of this specification, a computer-readable medium does not include a transient computer-readable medium such as a modulated data signal and a modulated carrier.

[0061] It should also be noted that the terms “comprising,” “comprising,” or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, product, or device comprising a series of elements not only comprises such elements but also comprises other elements not explicitly enumerated, or further comprises elements unique to such process, method, product, or device. Unless otherwise specified, an element limited by “...comprising” does not exclude other identical elements present within the process, method, product, or device comprising such element.

[0062] The foregoing description is merely an example of the present disclosure and is not intended to limit the present disclosure. To a person skilled in the art, various modifications and variations may be made to the present disclosure. Any modification, equivalent substitution, improvement, etc. made within the spirit and principles of the present disclosure falls within the scope of protection of the present disclosure.

Claims

Claim 1 A method for predicting the capacity of a power battery, comprising: a step of acquiring sample data of the power battery by means of an input unit; a step of dividing the sample data into multiple categories using a clustering algorithm, wherein each of the categories has a corresponding aging model and a feature identifier, wherein the aging model is obtained by the following steps: determining a fitting relationship within the aging model and determining parameters of the fitting relationship according to sample data of the aging model of the corresponding type, and wherein the feature identifier is used to identify the features of the sample data of the corresponding category; a step of acquiring battery state parameters of the power battery to be tested by means of an input unit; a step of determining an aging model adopted by the battery state parameters from a plurality of aging models by means of a matcher; and a step of inputting the battery state parameters into the adopted aging model to obtain a corresponding battery capacity by means of a calculator, wherein the clustering algorithm is a DBSCAN algorithm; and the clustering parameters include a radius of neighbors and a neighbor count threshold. Claim 2 A method according to claim 1, wherein the sample data of the power battery includes a plurality of sets of historical data for the same model of the power battery under actual vehicle driving conditions. Claim 3 The method of claim 1, wherein the step of dividing the sample data into multiple categories using the clustering algorithm comprises: a step of pre-setting the clustering algorithm and determining clustering parameters within the clustering algorithm; a step of dividing the sample data into core points or boundary points according to the clustering parameters; and a step of configuring categories according to the core points and dividing the sample data into multiple categories. Claim 4 delete Claim 5 The method of claim 1, wherein the feature identifier is a clustering center; and the step of determining the aging model adopted by the battery state parameter from the plurality of aging models comprises: the step of calculating the distance between the battery state parameter and the clustering center corresponding to each category; and the step of selecting the closest aging model as the aging model adopted by the battery state parameter. Claim 6 A method according to claim 1, wherein the fitting relationship includes polynomial fitting, neural network fitting, or regression tree fitting. Claim 7 A method according to claim 1, wherein the battery state parameters include at least two of current, voltage, temperature, state of charge, storage time, depth of discharge, and Coulomb efficiency. Claim 8 A device for predicting the capacity of a power battery, comprising: an input unit configured to acquire battery state parameters; a matching unit configured to match the battery state parameters to determine an aging model adopted by the battery state parameters from a plurality of aging models, wherein the plurality of aging models correspond one-to-one with various categories of sample data divided by a clustering algorithm, and each of the aging models includes a mapping relationship between the battery state parameters and the battery capacity; and a calculator configured to input the battery state parameters into the adopted aging model to obtain a corresponding battery capacity, wherein the clustering algorithm is a DBSCAN algorithm; and the clustering parameters include a radius of neighbors and a neighbor count threshold. Claim 9 A device for predicting the capacity of a power battery, comprising: at least one processor; and a memory connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the at least one processor executes instructions stored in the memory to implement a method for predicting the capacity of a power battery according to any one of claims 1 to 3 and claims 5 to 7. Claim 10 A computer-readable storage medium storing a computer program, wherein the program, when executed by a processor, implements a method for predicting the capacity of a power battery according to any one of claims 1 to 3 and claims 5 to 7.

Images