CN116593904A - Model training method and method for predicting battery SOH and battery RUL - Google Patents
Model training method and method for predicting battery SOH and battery RUL Download PDFInfo
- Publication number
- CN116593904A CN116593904A CN202310871063.3A CN202310871063A CN116593904A CN 116593904 A CN116593904 A CN 116593904A CN 202310871063 A CN202310871063 A CN 202310871063A CN 116593904 A CN116593904 A CN 116593904A
- Authority
- CN
- China
- Prior art keywords
- battery
- charging
- whale
- prediction model
- soh
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 365
- 238000012549 training Methods 0.000 title claims abstract description 98
- 230000008569 process Effects 0.000 claims abstract description 271
- 238000007600 charging Methods 0.000 claims abstract description 195
- 238000007599 discharging Methods 0.000 claims abstract description 149
- 241000283153 Cetacea Species 0.000 claims description 187
- 238000004422 calculation algorithm Methods 0.000 claims description 43
- 230000036541 health Effects 0.000 claims description 35
- 238000000605 extraction Methods 0.000 claims description 26
- 238000010277 constant-current charging Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 22
- 238000005457 optimization Methods 0.000 claims description 21
- 239000013598 vector Substances 0.000 claims description 19
- 238000010280 constant potential charging Methods 0.000 claims description 18
- 238000012804 iterative process Methods 0.000 claims description 18
- 239000002245 particle Substances 0.000 claims description 14
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 abstract description 5
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 239000010410 layer Substances 0.000 description 23
- 238000004590 computer program Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 6
- 230000003287 optical effect Effects 0.000 description 6
- 238000003491 array Methods 0.000 description 3
- 238000010219 correlation analysis Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000002093 peripheral effect Effects 0.000 description 3
- 230000000630 rising effect Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 210000004027 cell Anatomy 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Abstract
The disclosure provides a model training method of a battery SOH prediction model and a RUL prediction model, and a method for predicting the battery SOH and the battery RUL, and relates to the technical fields of artificial intelligence and battery management, wherein the method comprises the following steps: acquiring historical operation data corresponding to a plurality of charging and discharging processes of a sample battery and a corresponding first battery SOH; wherein, the charging and discharging processes have corresponding orders; extracting features of historical operation data corresponding to each charging and discharging process respectively to obtain key operation features of each charging and discharging process; training a battery SOH prediction model based on key operation characteristics of each charge and discharge process and a corresponding first battery SOH to obtain a trained battery SOH prediction model; based on the corresponding sequence of each charging and discharging process and the first battery SOH, training the battery RUL prediction model to obtain a trained battery RUL prediction model. Therefore, the prediction effect of the model, namely the accuracy and the reliability of the prediction result of the model can be improved.
Description
Technical Field
The present disclosure relates to the field of artificial intelligence and battery management technologies, and in particular, to a model training method for a battery SOH prediction model and a RUL prediction model, and a method for predicting a battery SOH and a battery RUL.
Background
Currently, batteries are widely used in energy storage technology due to their high energy density and recharging characteristics. However, as the battery is used, its performance may decay with increasing cycle times (i.e., the number of charge and discharge times). How to quickly and accurately estimate the SOH (State of Health) of a battery and predict the RUL (remaining life) of the battery, make a judgment on the performance of the battery for an auxiliary user, and improve the safety and reliability of the battery and the electric equipment is very important.
Disclosure of Invention
The present disclosure aims to solve, at least to some extent, one of the technical problems in the related art.
The present disclosure provides a model training method for a battery SOH prediction model and a RUL prediction model and a method for predicting a battery SOH and a battery RUL, so as to achieve training of the battery SOH prediction model and the battery RUL prediction model, and improve the prediction effect of the model, that is, improve the accuracy and reliability of the model prediction result.
An embodiment of a first aspect of the present disclosure provides a model training method for a battery SOH prediction model and a RUL prediction model, including:
acquiring historical operation data corresponding to a plurality of charging and discharging processes of a sample battery and corresponding first battery health states SOH; any one of the historical operation data comprises voltage information, current information and temperature information of the sample battery in the corresponding charging and discharging process; the charging and discharging processes have corresponding sequences;
Extracting features of historical operation data corresponding to each charging and discharging process respectively to obtain key operation features of each charging and discharging process;
based on key operation characteristics of each charging and discharging process and the corresponding first battery SOH, performing model training on the battery SOH prediction model to obtain a trained battery SOH prediction model; the battery SOH prediction model is an extreme learning machine ELM model;
model training is carried out on the battery RUL prediction model based on the sequence corresponding to the charging and discharging processes and the first battery SOH so as to obtain a trained battery RUL prediction model; the battery RUL prediction model is a model constructed based on a particle filter PF algorithm.
Embodiments of a second aspect of the present disclosure propose a method for predicting a battery SOH and a battery RUL, comprising:
obtaining a trained battery state of health SOH prediction model and a battery residual service life RUL prediction model by adopting a model training method of the battery SOH prediction model and the RUL prediction model provided by the embodiment of the first aspect;
acquiring target operation data of a target charging and discharging process of a target battery;
extracting features of the target operation data to obtain target key operation features of the target charging and discharging process;
Based on the target key operation characteristics, predicting the battery SOH of the target battery by adopting the trained battery SOH prediction model so as to obtain the target battery SOH;
based on the target battery SOH, predicting the sequence of the target charge-discharge process by adopting the trained battery RUL prediction model so as to obtain a target sequence;
and determining a target battery RUL of the target battery according to the set life threshold of the target battery and the target sequence.
An embodiment of a third aspect of the present disclosure provides a model training apparatus for a battery SOH prediction model and a RUL prediction model, including:
the acquisition module is used for acquiring historical operation data corresponding to a plurality of charging and discharging processes of the sample battery and corresponding first battery health states SOH; any one of the historical operation data comprises voltage information, current information and temperature information of the sample battery in the corresponding charging and discharging process; the charging and discharging processes have corresponding sequences;
the extraction module is used for extracting the characteristics of the historical operation data corresponding to each charging and discharging process respectively so as to obtain the key operation characteristics of each charging and discharging process;
The first training module is used for carrying out model training on the battery SOH prediction model based on key operation characteristics of each charging and discharging process and the corresponding first battery SOH so as to obtain a trained battery SOH prediction model; the battery SOH prediction model is an extreme learning machine ELM model;
the second training module is used for carrying out model training on the battery RUL prediction model based on the sequence corresponding to the charging and discharging processes and the first battery SOH so as to obtain a trained battery RUL prediction model; the battery RUL prediction model is a model constructed based on a particle filter PF algorithm.
An embodiment of a fourth aspect of the present disclosure proposes an apparatus for predicting a battery SOH and a battery RUL, comprising:
the first obtaining module is used for obtaining a trained battery state of health (SOH) prediction model and a battery residual service life (RUL) prediction model by adopting a model training method of the SOH prediction model and the RUL prediction model provided by the embodiment of the first aspect;
the second acquisition module is used for acquiring target operation data of a target charging and discharging process of the target battery;
the extraction module is used for extracting the characteristics of the target operation data to obtain the target key operation characteristics of the target charging and discharging process;
The first prediction module is used for predicting the battery SOH of the target battery by adopting the trained battery SOH prediction model based on the target key operation characteristics so as to obtain the target battery SOH;
the second prediction module is used for predicting the sequence of the target charge and discharge process by adopting the trained battery RUL prediction model based on the target battery SOH so as to obtain a target sequence;
and the determining module is used for determining a target battery RUL of the target battery according to the set life threshold of the target battery and the target sequence.
An embodiment of a fifth aspect of the present disclosure proposes an electronic device, including: the system comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the model training method of the battery SOH prediction model and the RUL prediction model as set forth in the embodiment of the first aspect of the disclosure or the method for predicting the battery SOH and the battery RUL as set forth in the embodiment of the second aspect of the disclosure.
An embodiment of a sixth aspect of the present disclosure proposes a non-transitory computer-readable storage medium, on which a computer program is stored, which when executed by a processor, implements a model training method for predicting a battery SOH and a battery RUL as proposed by the embodiment of the first aspect of the present disclosure, or a method for predicting a battery SOH and a battery RUL as proposed by the embodiment of the second aspect of the present disclosure.
An embodiment of a seventh aspect of the present disclosure proposes a computer program product, which when executed by a processor, performs a model training method of a battery SOH prediction model and a RUL prediction model as proposed by an embodiment of the first aspect of the present disclosure, or a method for predicting a battery SOH and a battery RUL as proposed by an embodiment of the second aspect of the present disclosure.
Additional aspects and advantages of the disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the disclosure.
Drawings
The foregoing and/or additional aspects and advantages of the present disclosure will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a flowchart of a model training method for SOH and RUL prediction models of a battery according to an embodiment of the present disclosure;
fig. 2 is a schematic diagram of a voltage signal curve of a constant current charging stage of a sample battery provided by the present disclosure;
fig. 3 is a schematic diagram of a current signal curve of a constant voltage charging stage of a sample battery provided by the present disclosure;
fig. 4 is a flowchart of a model training method of a battery SOH prediction model and a RUL prediction model according to a second embodiment of the disclosure;
FIG. 5 is a flow chart of a method for predicting a battery state of health SOH and a battery remaining life RUL according to a third embodiment of the present disclosure;
fig. 6 is a schematic structural diagram of a model training device for a battery SOH prediction model and a RUL prediction model according to a fourth embodiment of the present disclosure;
fig. 7 is a schematic structural diagram of an apparatus for predicting a battery state of health SOH and a remaining battery life RUL according to a fifth embodiment of the present disclosure;
FIG. 8 illustrates a schematic block diagram of an example electronic device that may be used to implement embodiments of the present disclosure.
Detailed Description
Embodiments of the present disclosure are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and intended for the purpose of explaining the present disclosure and are not to be construed as limiting the present disclosure.
The battery SOH estimation or prediction aims to evaluate the current state of health of the battery to determine its capacity and degree of power decay; battery RUL prediction aims to predict the time that a battery may last in the future. Currently, battery SOH estimation and battery RUL prediction can be achieved through various methods, such as a statistical analysis method based on real-time performance data and historical data of a battery, a whale optimization algorithm, a pearson correlation coefficient and gray correlation analysis method, a machine learning algorithm, and the like.
Although the above-described methods have good effects on battery SOH estimation and battery RUL prediction, some methods have room for improvement, mainly in the following aspects:
(1) The whale optimization algorithm is widely applied to the hyper-parameter optimization of the reference function test and the prediction model. However, the standard whale optimization algorithm still has the defects of low convergence speed and low precision in the super-parameter optimization process of machine learning or deep learning.
(2) The pearson correlation coefficient and gray correlation analysis method can reduce the dimension of input features and improve the learning capacity and generalization capacity of the prediction model. However, pearson correlation coefficients and gray correlation analysis do not take into account the interactions of the various features on the prediction model nor do it analyze the importance of the prediction targets.
Random Forest (RF) and gradient-enhanced regression trees use a fixed number of feature subsets for optimal segmentation. But sometimes the target contribution value returned by multiple features is 0, which is not effective in ranking the contributions of all features.
(3) Some of the methods described above extract features with high correlation from the IC curves and estimate the battery SOH by data fitting. However, these methods require predictive tag data in advance to obtain good data fitting parameters. When the charge and discharge strategies are inconsistent, the parameters identified by one battery are difficult to apply to SOH estimation of the other battery. Although the extremum learner algorithm can realize training and estimation among different battery SOHs, has higher robustness and universality, the extremum learner algorithm does not consider the influence of new samples and old samples on model prediction performance, so that the extremum learner algorithm has weaker adaptability to complex time-varying systems.
In view of at least one of the above problems, the present disclosure proposes a model training method of a battery SOH prediction model and a RUL prediction model, and a method for predicting a battery SOH and a battery RUL.
The model training method of the battery SOH prediction model and the RUL prediction model and the method for predicting the battery SOH and the battery RUL according to the embodiments of the present disclosure are described below with reference to the accompanying drawings.
Fig. 1 is a flowchart of a model training method of a battery SOH prediction model and a RUL prediction model according to an embodiment of the disclosure.
The embodiment of the disclosure is exemplified by the model training method of the battery SOH prediction model and the RUL prediction model being configured in the model training device of the battery SOH prediction model and the RUL prediction model, and the model training device of the battery SOH prediction model and the RUL prediction model can be applied to any electronic device so that the electronic device can execute the model training functions of the battery SOH prediction model and the RUL prediction model.
The electronic device may be any device with computing capability, for example, may be a personal computer (Personal Computer, abbreviated as PC), a mobile terminal, a server, etc., and the mobile terminal may be a mobile phone, a tablet computer, a personal digital assistant, a wearable device, etc. with various operating systems, a touch screen, and/or a hardware device with a display screen.
As shown in fig. 1, the model training method of the battery SOH prediction model and the RUL prediction model may include the steps of:
step 101, obtaining historical operation data corresponding to a plurality of charging and discharging processes of a sample battery and corresponding first battery health states SOH.
Any historical operation data may include voltage information, current information, temperature information and the like of the sample battery in the corresponding charging and discharging processes, which is not limited in the present disclosure.
Wherein the charging and discharging processes may have a corresponding order. For example, a certain charge-discharge process of the sample battery is an i-th charge-discharge process of the sample battery, and the corresponding sequence of the charge-discharge processes is i, where i is a positive integer greater than 0.
In the embodiment of the present disclosure, each charge-discharge process of the sample battery may have a corresponding battery state of health SOH, which is denoted as the first battery SOH in the present disclosure.
In the embodiment of the disclosure, historical operation data corresponding to a plurality of charging and discharging processes of a sample battery and corresponding first battery health states can be obtained.
As an example, a charge and discharge experiment may be performed on a sample battery in advance, historical operation data corresponding to a plurality of charge and discharge processes of the sample battery may be recorded, and the first battery SOH corresponding to each charge and discharge process may be measured offline.
In one possible implementation of the embodiments of the present disclosure, the charging and discharging process may include a charging process, and the charging process may include a constant current charging stage and a constant voltage charging stage.
And 102, respectively extracting features of historical operation data corresponding to each charging and discharging process to obtain key operation features of each charging and discharging process.
In the embodiment of the disclosure, feature extraction can be performed on the historical operation data corresponding to each charging and discharging process respectively to obtain key operation features of each charging and discharging process, that is, for any charging and discharging process, feature extraction can be performed on the historical operation data corresponding to the charging and discharging process to obtain key operation features of the charging and discharging process.
As a possible implementation manner, feature extraction can be performed on the historical operation data corresponding to each charging and discharging process respectively to obtain initial operation features of each charging and discharging process; wherein the initial operating characteristics may include voltage characteristics, current characteristics, temperature characteristics, and IC (Incremental Capacity, delta capacity) curve characteristics; for any charge and discharge process, feature analysis based on an extreme random tree algorithm can be performed on the initial operation features of the charge and discharge process, so that the corresponding key operation features are screened and reserved from the initial operation features of the charge and discharge process.
In the disclosed embodiments, voltage characteristics may be used to indicate voltage related characteristics.
In the disclosed embodiments, the current signature may be used to indicate a current-related signature.
In the disclosed embodiments, temperature characteristics may be used to indicate temperature related characteristics.
In the disclosed embodiments, the IC curve features may be used to indicate IC curve related features.
In the embodiment of the disclosure, for any charge and discharge process, feature extraction can be performed on historical operation data corresponding to the charge and discharge process, so as to obtain initial operation features of the charge and discharge process.
For clarity of explanation, for any charge and discharge process, how to perform feature extraction on historical operation data corresponding to the charge and discharge process to obtain initial operation features of the charge and discharge process, in a possible implementation manner of the embodiment of the disclosure, when the charge and discharge process includes a charge process, the charge process includes a constant current charge stage and a constant voltage charge stage, for any charge and discharge process, the following steps may be adopted to obtain the initial operation features of the charge and discharge process:
1. the characteristic extraction can be carried out on the voltage information corresponding to the constant current charging stage in the charging and discharging process, so that the statistical characteristic of the extracted voltage information is used as the voltage characteristic in the charging and discharging process.
It should be noted that the statistical features of the voltage information are not limited in this disclosure, for example, the statistical features of the voltage information may include a time interval of equal charging voltage rise, a charging capacity of equal charging voltage rise, a slope of a charging voltage curve, and the like.
The time interval of the rising of the constant charging voltage may refer to a time length between a start time of the constant current charging stage and any time of the constant current charging stage that the voltage signal becomes stable after rising.
The charging capacity of the equal charging voltage rise may refer to a capacitance increment corresponding to a voltage signal rising stage of the constant current charging stage under the condition of the same charging voltage in different charging and discharging processes.
The slope of the charging voltage curve may refer to a total duration between a start time and a stop time of the constant current charging stage, and a ratio of absolute values of differences between the start voltage corresponding to the start time and the stop voltage corresponding to the stop time of the constant current charging stage. It should be noted that, the time interval of the equal charging voltage rise may be equal to the total duration between the start time and the stop time of the constant current charging stage, or may not be equal to the total duration between the start time and the stop time of the constant current charging stage, which is not limited in the present disclosure.
As an example, when extracting characteristics of voltage information corresponding to a constant current charging stage in a charging and discharging process of a sample battery, a voltage signal curve of the constant current charging stage of the sample battery is shown in fig. 2, and when a voltage in the charging process reaches a set voltage threshold (such as 3.2V, 3.3V, etc.), the sample battery enters the constant current charging stage, and a mark U is marked 0 For the initial voltage corresponding to the initial time of the constant current charging stage, a mark U is marked k For the cut-off voltage corresponding to the cut-off time of the constant current charging stage, the voltage of the sample battery can be changed from the voltage to U in the constant current charging stage 0 To a voltage of U k Time interval deltat of (a) v Slope K of charging voltage curve as time interval of equal charging voltage rise v Can be expressed as:
；（1）
in one possible implementation of the embodiments of the present disclosure, where the statistical characteristic of the voltage information includes a charging capacity of an equal charging voltage rise, the charging capacity of the equal charging voltage rise may be obtained based on a PF (Particle Filtering, particle filter) algorithm.
As an example, assume the state equation and the observation equation of the nonlinear system employed by the particle filtering algorithm are:
；（2）
wherein ,xk Is a state vector, in the present disclosure, the state vector may represent a battery state of Charge (SOC), and in practical application, the SOC cannot be directly measured and obtained, and a particle filter is required to be used to estimate the SOC; z is Z k Is the observation vector, Z in this disclosure k The battery terminal voltage can be indicated and can be obtained through sensor measurement; omega k-1 and vk Process noise and measurement noise, respectively. The subscript k characterizes the current instant and k-1 characterizes the instant immediately preceding the instant k.
By adopting the method, the value x of the SOC of the sample battery at the charging starting moment of the constant-current charging stage can be estimated 0 And the value x of SOC at the charge cutoff time n And the charging capacity C of the equal charging voltage rise of the sample battery in the constant current charging stage can be determined according to the following formula:
；（3）
where Q is the rated capacity of the sample cell.
Therefore, in the present disclosure, feature extraction can be performed on the voltage information corresponding to the constant current charging stage in the charging and discharging process, so as to extract the statistical feature of the voltage information, and after the statistical feature of the extracted voltage information, the statistical feature of the extracted voltage information can be used as the voltage feature in the charging and discharging process.
2. The current information corresponding to the constant voltage charging stage in the charging and discharging process can be subjected to feature extraction, so that the statistical features of the extracted current information are used as the current features in the charging and discharging process.
It should be noted that, the statistical features of the current information are not limited in this disclosure, for example, the statistical features of the current information may include a time interval of an equal charge current decrease, a charge capacity of the equal charge current decrease, a slope of a charge current curve, and the like.
The time interval of the equal charging current drop may refer to a time length between a start time of the constant voltage charging stage and any time of the current signal drop stage in the same charging current in different charging and discharging processes.
The charging capacity of the equal charging current drop may refer to a capacitance increment corresponding to a current signal drop stage in a constant voltage charging stage under the same charging current in different charging and discharging processes.
The slope of the charging current curve may refer to a ratio of an absolute value of a difference between a start current corresponding to the start time and an off current corresponding to any time, for a time length between the start time of the constant voltage charging stage and any time of the constant voltage charging stage current signal falling stage.
It should be noted that, the time interval of the equal charging current drop may be equal to the total duration between the start time and the off time of the constant voltage charging stage, or may not be equal to the total duration between the start time and the off time of the constant voltage charging stage, which is not limited in the present disclosure.
As an example, when extracting the characteristics of the current information corresponding to the constant voltage charging stage in the charge and discharge process of the sample battery, the current signal curve of the constant voltage charging stage of the sample battery is shown in fig. 3, it can be understood that the charging process of the sample battery enters the constant voltage charging stage first and then enters the constant current charging stage, the initial time of the constant voltage charging stage is 0, the current corresponding to the initial time is marked as I, and any time of the current signal falling stage of the constant voltage charging stage is marked as t K The current corresponding to any time is I K The sample battery can be charged from the current I in the constant voltage stage 0 To a current of I K Time interval deltat of (a) I （=t K ) Slope K of charging current curve as time interval of equal charging current up and down I Can be expressed as:
；（4）
therefore, in the present disclosure, the current information corresponding to the constant voltage charging stage in the charging and discharging process may be extracted to extract the statistical feature of the current information, and the statistical feature of the extracted current information may be used as the current feature in the charging and discharging process after the statistical feature of the extracted current information.
3. The characteristic extraction can be performed on the temperature information corresponding to the charging process in the charging and discharging process, so that the statistical characteristic of the extracted temperature information is used as the temperature characteristic of the charging and discharging process.
It should be noted that, the statistical features of the temperature information are not limited in this disclosure, for example, the statistical features of the temperature information may include an average surface temperature of the battery, an area under a temperature curve of the battery, a maximum surface temperature of the battery, and a time point corresponding to the maximum surface temperature of the battery.
Wherein, the average surface temperature of the battery refers to the average value of the surface temperature of the battery during charging.
The area under the battery temperature curve refers to the area of an area surrounded by the battery temperature curve and a corresponding coordinate axis in the charging process.
Wherein, the maximum surface temperature area of the battery refers to the maximum value of the surface temperature of the battery during the charging process.
In the embodiment of the disclosure, the feature extraction can be performed on the temperature information corresponding to the charging process in the charging and discharging process, so as to extract the statistical feature of the temperature information, and the statistical feature of the extracted temperature information can be used as the temperature feature of the charging and discharging process.
According to the current information and the voltage information corresponding to the charging process in the charging and discharging processes, an IC curve can be generated. It can be understood that the corresponding IC curves can be generated according to the current information and the voltage information corresponding to the charging process in the charging and discharging processes.
It should be noted that, in the battery pack, due to the inconsistency between the battery cells, generating the corresponding IC curve according to the current information and the voltage information corresponding to the charging process in the charging and discharging process generally includes a large amount of noise, and therefore, in one possible implementation manner of the embodiment of the disclosure, a gaussian filter may be used to smooth the IC curve to obtain the processed IC curve. Thus, the accuracy and definition of the IC curve can be improved.
5. The IC curve may be subjected to feature extraction to take the statistical features of the extracted IC curve as IC curve features of the charge-discharge process.
It should be noted that the present disclosure does not limit the statistical characteristics of the IC curve. For example, the statistical features of the IC curve may include the peak value, peak position, area under the IC curve, and the like of the IC curve.
The area under the IC curve may indicate the area of the area surrounded by the IC curve and the corresponding coordinate axis.
In the embodiment of the disclosure, the characteristic extraction may be performed on the IC curve to extract the statistical characteristic of the IC curve, and the extracted statistical characteristic of the IC curve may be used as the IC curve characteristic of the charging and discharging process.
Therefore, the characteristic extraction can be carried out on the historical operation data corresponding to the charging and discharging process based on the historical operation data corresponding to the charging and discharging process, and the initial operation characteristics of the charging and discharging process can be effectively obtained.
In the embodiment of the disclosure, for any charge and discharge process, feature analysis based on an extreme random tree algorithm can be performed on initial operation features of the charge and discharge process, so as to screen and retain corresponding key operation features from the initial operation features of the charge and discharge process.
As an example, an extreme random number algorithm may be used to determine the contribution of each of the initial operating characteristics of the charge-discharge process; secondly, sequencing all the features in the initial operation features of the charge and discharge process according to the sequence from the large contribution value to the small contribution value so as to obtain a sequencing sequence; finally, each feature in the sorting sequence, which is not more than the set order, can be used as a key operation feature of the charging and discharging process. It should be noted that, the present disclosure does not limit the value of the setting order, for example, the setting order may be 5, 6, etc., and the setting order may be set according to actual needs.
Therefore, the initial operation characteristics based on the charging and discharging process can be realized, and the key operation characteristics of the charging and discharging process can be effectively obtained.
And 103, performing model training on the battery SOH prediction model based on key operation characteristics of each charge and discharge process and the corresponding first battery SOH to obtain a trained battery SOH prediction model.
The battery SOH prediction model may be an ELM (Extreme Learning Machine ) model, among others.
In the embodiment of the disclosure, the battery SOH prediction model may be model-trained based on the key operation characteristics of each charge and discharge process and the corresponding first battery SOH, so that the battery SOH prediction model may learn the correspondence between the key operation characteristics of the charge and discharge process of the battery and the battery SOH, and thus a trained battery SOH prediction model may be obtained.
In order to clearly illustrate how the battery SOH prediction model is model trained, the above process is described in detail below with reference to examples. Firstly, it should be noted that the ELM model is a single-layer feedforward neural network, has a strong nonlinear approximation capability, and has the characteristics of fast learning speed and less calculation amount. As shown in fig. 4, fig. 4 is a network structure of an ELM model, which can be expressed as follows:
；（5）
Wherein, gelm (·) is a Sigmoid activation function;is the input weight of the ith hidden layer node in the network corresponding to the ELM model, i is E [1, L]Wherein L is the total number of hidden layer nodes of the network corresponding to the ELM model; />Is the bias coefficient of the i-th hidden layer node in the network corresponding to the ELM model; />Is the output weight connecting the i hidden layer node and the output layer node; />Is the input value of the network corresponding to the ELM model; />Is the output value of the network corresponding to the ELM model; j E [1, N]Wherein N is a training sampleTotal number.
The goal of the network correspondence for ELM models is to minimize the output error, and its objective function can be expressed as follows:
；（6）
the network to which the ELM model corresponds may translate the problem into a matrix form as shown in equations (7) and (8):
；（7）
；（8）
wherein ,Helm Is the output matrix of the hidden layer, beta elm Is made up of beta i （i∈[1,L]) Matrix of Y is composed of Y j （j∈[1,N]) A matrix is formed.
The training process of the ELM model ultimately converts the problem into a solution to the minimum norm of the linear system (7), and the output weights connecting the hidden layer node and the output layer node can be determined by the following formula:
；（9）
wherein ,is a matrix->Is a generalized inverse matrix of (a).
According to the above ELM model training process, in the present disclosure, based on key operation features of each charge and discharge process and the corresponding first battery SOH, a recursive algorithm may be used to perform model training on a battery SOH prediction model, which may include the following two phases:
(1) Initialization phase
Assume that the number of charge and discharge processes of the sample battery is N 0 The method comprises the steps of carrying out a first treatment on the surface of the The key operation characteristic of the jth charge-discharge process is x j The first battery SOH in the jth charge-discharge process is y j ，j∈[1,N 0 ]Taking key operation characteristics of the charge and discharge processes as input values of a network corresponding to the ELM model, taking a first battery SOH of each charge and discharge process as a corresponding output value of the network corresponding to the ELM model, and connecting output weights of the hidden layer node and the output layer nodeInitial value +.>The following formula is used for determination:
；（10）
；（11）
；（12）
；（13）
(2) Online learning stage:
based on the new data set of the k+1th block, online updating the output weight by adopting a recursion method:
；（14）
where k is the order in the total number of successive iterations selected when a regression algorithm is employed, t k+1 The iteration number in the model training process is represented by I, which is an identity matrix.
In order to improve the prediction performance of the ELM model, in one possible implementation of the embodiments of the present disclosure, a variable forgetting factor is used in the ELM model to adjust the output weights connecting each hidden layer node and each output layer node in the ELM model.
Still described in the above examples, a variable forgetting factorCan be expressed as follows:
；（15）
；（16）
wherein ,A predetermined minimum forgetting factor; />May be a preset root mean square error threshold.
Variable forgetting factor determined using equations (15) and (16)After that, the hidden layer node and the output layer node are connected with the output weight +.>Can be represented by the equation (14) changed to the equation (17):
；（17）
and 104, performing model training on the battery RUL prediction model based on the sequence corresponding to each charging and discharging process and the first battery SOH so as to obtain a trained battery RUL prediction model.
The battery RUL prediction model may be a model constructed based on a particle filter PF algorithm.
As an example, the battery RUL prediction model constructed based on the particle filtering PF algorithm may be a double-exponential function model, where the state equation of the double-exponential function in the model is formula (18), and the measurement equation is formula (19):
；（18）
；（19）
wherein ,，/>is a state vector, a p,n 、b p,n 、c p,n 、d p,n Model parameters of a battery RUL prediction model; v n ∈N(0,σ v )，ω a,n-1 、ω b,n-1 、ω c,n-1 、ω d,n-1 and vn All are noise with average value of 0, sigma a 、σ b 、σ c 、σ d and σv All are variances of noise; n is the corresponding sequence of the battery charging and discharging process; />And predicting the battery SOH corresponding to the nth charge and discharge process of the obtained battery.
In the embodiment of the disclosure, the model training may be performed on the battery RUL prediction model based on the sequence corresponding to each charging and discharging process and the first battery SOH, so that the battery RUL prediction model may learn the correspondence between the sequence of the charging and discharging processes of the battery and the battery SOH, and thus a trained battery RUL prediction model may be obtained.
According to the model training method of the battery SOH prediction model and the RUL prediction model, historical operation data corresponding to a plurality of charging and discharging processes of a sample battery and corresponding first battery health state SOH are obtained; any historical operation data comprises voltage information, current information and temperature information of the sample battery in the corresponding charging and discharging processes; the charging and discharging processes have corresponding sequences; extracting features of historical operation data corresponding to each charging and discharging process respectively to obtain key operation features of each charging and discharging process; based on key operation characteristics of each charge and discharge process and the corresponding first battery SOH, performing model training on the battery SOH prediction model to obtain a trained battery SOH prediction model; the battery SOH prediction model is an extreme learning machine ELM model; model training is carried out on the battery RUL prediction model based on the corresponding sequence of each charging and discharging process and the first battery SOH so as to obtain a trained battery RUL prediction model; the battery RUL prediction model is a model constructed based on a particle filter PF algorithm. Therefore, the battery SOH prediction model and the battery RUL prediction model can be trained, the prediction effect of the model is improved, and the accuracy and reliability of the model prediction result are improved.
In order to clearly illustrate how to model-train a battery SOH prediction model based on key operation characteristics of each charge-discharge process and a corresponding first battery SOH in the above embodiments of the present disclosure, the present disclosure further provides a model training method of the battery SOH prediction model and the RUL prediction model.
Fig. 4 is a flowchart of a model training method of a battery SOH prediction model and a RUL prediction model according to a second embodiment of the disclosure.
As shown in fig. 4, the model training method of the battery SOH prediction model and the RUL prediction model may include the steps of:
step 401, obtaining historical operation data corresponding to a plurality of charging and discharging processes of the sample battery and corresponding first battery state of health SOH.
And step 402, respectively extracting features of historical operation data corresponding to each charging and discharging process to obtain key operation features of each charging and discharging process.
The execution of steps 401 to 402 may refer to the execution of any embodiment of the disclosure, and will not be described herein.
Step 403, performing model training based on a whale optimization algorithm on the battery SOH prediction model based on key operation characteristics of each charge and discharge process and the corresponding first battery SOH, so as to obtain a trained battery SOH prediction model.
In the embodiment of the disclosure, based on key operation characteristics of each charge and discharge process and the corresponding first battery SOH, model training based on a whale optimization algorithm can be performed on the battery SOH prediction model, and a trained battery SOH prediction model can be obtained.
As one possible implementation, the initial value of the total number of whales and the position value of each whale of the whale population may be set to generate a whale initial population; the position value of the whale can represent the value of the super parameter of the battery SOH prediction model; the battery SOH prediction model can be subjected to an iteration process which is not more than a set iteration number and is based on a whale optimization algorithm based on key operation characteristics of each charging and discharging process, a corresponding first battery SOH and a whale initial population, so that optimal whales in the whale population can be updated based on fitness values of each whale in any iteration process.
The super parameter of the battery SOH prediction model may be, for example, the number of hidden layers, the number of neurons of each layer, and the like, which is not limited in the disclosure.
The set number of iterations may be preset, for example, may be 5000, 10000, etc., which is not limited in this disclosure.
Wherein the fitness value of the optimal whale can be no more than the fitness value of any whale except the optimal whale in the whale population; the fitness value of any whale can be determined by inputting key operation characteristics of each charging and discharging process into a battery SOH prediction model adopting position values of corresponding whales and adopting an objective function according to output of the battery SOH prediction model and first battery SOH corresponding to each charging and discharging process.
As an example, assume that the total number of whales in the whale population is 5, and that whale A is present therein, the total number of the plurality of charge and discharge processes is N, and the key operating characteristic of the jth charge and discharge process is x j ，j∈[1,N]And j is a positive integer, the key operating feature of the battery SOH prediction model (i.e., ELM model) that uses the position value of whale A for input is x j The output y corresponding to the battery SOH prediction model can be determined according to the following formula j ：
；（20）
wherein ,is a Sigmoid activation function; />Is the input weight of the ith hidden layer node in the network corresponding to the ELM model, i is E [1, L]Wherein L is the total number of hidden layer nodes of the network corresponding to the ELM model; />Is the bias coefficient of the i-th hidden layer node in the network corresponding to the ELM model; />Is the output weight connecting the i-th hidden layer node and the output layer node.
In connection with equation (20), the objective function may be expressed as follows:
；（21）
wherein ,and the first battery SOH corresponding to the jth charging and discharging process.
From equation (20) and equation (21), the Fitness value Fitness of whale A can be determined A The method comprises the following steps:
；（22）
from this, the fitness value of whale a can be determined.
Similar to determining the fitness value of whale a, the fitness value of other whales in the whale population may also be determined, and will not be described in detail herein.
In each iteration of updating the optimal whale in the whale population based on the fitness value of each whale, the optimal whale of the present round may be determined according to the fitness value of each whale, wherein the fitness value of the optimal whale is not greater than the fitness value of any whale in the whale population other than the optimal whale. For example, in an iteration process in which an optimal whale in a whale population is updated, when the fitness value of whale A is not greater than the fitness value of other whales in the whale population except whale A, whale A can be used as the optimal whale in the iteration process of the round.
It should be noted that the above total number of whales in the whale population is merely exemplary, and in practical application, the total number of whales in the whale population may be set as needed, which is not limited in this disclosure.
To clearly illustrate how the optimal whale in the whale population is updated based on the fitness value of each whale during any one iteration, in one possible implementation of the disclosed embodiments, the following steps may be taken when updating the optimal whale in the whale population based on the fitness value of each whale:
1. for any round of iterative process, under the condition that the sequence of the iterative process is not more than the set iterative times, the position value and the fitness value of the optimal whale of the previous round can be obtained.
As an example, for the t+1st round of iterative process, in the case that the order (t+1) of the iterative process is not greater than the set number of iterations, the position value and fitness value of the optimal whale in the last round, that is, the position value and fitness value of the optimal whale in the t-th round, may be obtained, where t is a positive integer.
2. The position of any whale in the whale population can be updated using the following formula, so that the position value of the whale in the whale population after updating is used as the position value of the whale in the whale population in the present round:
；（23）
；（24）
wherein ,；（25）
；（26）
；（27）
；（28）
；（29）
wherein t is the sequence of the previous iteration process, and t+1 is the sequence of the present iteration process; the number of whales in the whale population is N; x is X woa (. Cndot.) is the position vector corresponding to the whale population of the corresponding iterative process, wherein X woa Any element in (-) is a position value corresponding to whale; x is X rand (t) is a randomly generated position vector corresponding to the whale population of the previous round, wherein X rand Any one of the elements (t) is a randomly generated position value for the corresponding whale;is a vector of size 1*N, and +.>The value of any element is equal to the position value of the optimal whale in the round; a is that woa(·) and Cwoa (. Cndot.) are all variable coefficients; d (D) woa (. Cndot.) represents the distance of movement; r is (r) woa (. Cndot.) is that the elements are all 0,1]A vector of random numbers in (a); b is a vector with elements of 1; t is t max To set the iteration number, b woa Is a constant equal to 1, l woa Is [0,1 ]]Random number, p woa (. Cndot.) is [0,1]N (0, 1) represents a gaussian function with an average value of 0 and a variance of 1;
3. and determining the fitness value of the whale of the round based on the position value of the whale aiming at any whale of the whale population of the round.
That is, for any whale in the whale population of the present round, key operation characteristics of each charge and discharge process are input into a battery SOH prediction model adopting the position value of the whale, and the fitness value of the whale of the present round is determined by adopting an objective function according to the output of the battery SOH prediction model and the first battery SOH corresponding to each charge and discharge process.
4. And determining the optimal whale of the round according to the fitness value of each whale of the round and the fitness value of the optimal whale of the previous round.
As an example, the fitness value of each whale of the present round may be compared with the fitness value of the best whale of the previous round, and when the fitness value of a whale B in each whale of the present round is not greater than the fitness value of the best whale of the previous round, and the fitness value of the whale B of the present round is not greater than the fitness value of any whale other than the whale in the present round, it may be determined that the whale B is the best whale of the present round.
Thus, an update of the optimal whale in the whale population during any one iteration can be achieved.
In another possible implementation manner of the embodiment of the present disclosure, when the order of the iterative process is greater than the set number of iterations, the position value of the last round of optimal whale may be used as the optimal solution of the super parameter of the battery SOH prediction model, and the iterative process of this round may be stopped.
Therefore, the optimal solution of the super parameter of the battery SOH prediction model can be effectively obtained, and the training of the battery SOH prediction model is realized.
And step 404, performing model training on the battery RUL prediction model based on the sequence corresponding to each charging and discharging process and the first battery SOH to obtain a trained battery RUL prediction model.
The implementation of step 404 may refer to the implementation of any embodiment of the disclosure, which is not described herein.
It will be appreciated that model parameters, such as the weight of each particle, exist in the battery RUL prediction model (i.e., the model constructed based on the particle filter PF algorithm), and the above whale optimization algorithm may also be used to optimize the model parameters in the battery RUL prediction model. Thus, in one possible implementation of the embodiments of the present disclosure, model training based on a whale optimization algorithm may be performed on the battery RUL prediction model based on the sequence corresponding to each charge-discharge process and the first battery SOH to obtain a trained battery RUL prediction model.
That is, similar to the model training of the whale optimization algorithm for the battery SOH prediction model, the model training based on the whale optimization algorithm may also be performed for the battery RUL prediction model, so that a trained battery RUL prediction model may be obtained, which will not be described herein.
It should be noted that the above examples of the model parameters of the battery RUL prediction model are merely exemplary, and in practical applications, the model parameters of the battery RUL prediction model may also include other model parameters, and the present disclosure does not limit the model parameters of the battery RUL prediction model.
According to the model training method for the battery SOH prediction model and the RUL prediction model, model training based on a whale optimization algorithm is conducted on the battery SOH prediction model based on key operation characteristics of each charging and discharging process and corresponding first battery SOH, so that a trained battery SOH prediction model is obtained. Therefore, the model training of the battery SOH prediction model can be realized based on a whale optimization algorithm, the convergence rate of the model is accelerated, the global optimizing capability of the model is improved, and the model identification capability can be improved.
The above-described embodiments correspond to the model training methods of the battery SOH prediction model and the RUL prediction model, and the following embodiments apply the trained battery SOH prediction model and RUL prediction model.
Fig. 5 is a flowchart of a method for predicting a battery SOH and a battery RUL according to a third embodiment of the present disclosure.
As shown in fig. 5, the method for predicting the battery SOH and the battery RUL may include the steps of:
step 501, a battery SOH prediction model and a battery RUL prediction model are obtained through a model training method of the battery SOH prediction model and the RUL prediction model.
In the embodiment of the present disclosure, the battery SOH prediction model and the RUL prediction model may be obtained by using the model training method of the battery SOH prediction model and the RUL prediction model of any of the above embodiments.
Step 502, obtaining target operation data of a target charge-discharge process of a target battery.
In the embodiment of the disclosure, the target battery may be a battery to be predicted, and the target charge-discharge process of the target battery may be a charge-discharge process of the battery SOH and the battery RUL corresponding to the battery to be predicted.
In the embodiment of the present disclosure, the target operation data may include voltage information, current information, temperature information, etc. of the target battery during the target charge and discharge, which is not limited by the present disclosure.
In the embodiment of the disclosure, the target operation data of the target charge and discharge process of the target battery can be acquired.
In step 503, feature extraction is performed on the target operation data to obtain the target key operation feature of the target charging and discharging process.
In the embodiment of the disclosure, feature extraction may be performed on the target operation data to obtain the target key operation feature of the target charging and discharging process.
It should be noted that, in this embodiment, the method for obtaining the target key operation characteristics of the target charge and discharge process of the target battery is similar to the method for obtaining the key operation characteristics of each charge and discharge process of the sample battery in step 102, and will not be described herein.
Step 504, based on the target key operation characteristics, predicting the battery SOH of the target battery by using the trained battery SOH prediction model to obtain the target battery SOH.
It will be appreciated that the trained battery SOH prediction model has learned the correspondence between key operating characteristics of the battery and the battery SOH. Therefore, in the method, the target key operation characteristics can be input into a trained battery SOH prediction model so as to predict the battery SOH of the target battery and obtain the target battery SOH.
Step 505, based on the target battery SOH, predicting the sequence of the target charge and discharge process by using the trained battery RUL prediction model, so as to obtain the target sequence.
It will be appreciated that the trained battery RUL prediction model has learned the correspondence between the sequence of battery charge-discharge processes and battery SOH. Therefore, in the present disclosure, after the target battery SOH of the target battery is obtained, the sequence of the target charge and discharge process may be predicted by using the trained battery RUL prediction model according to the target battery SOH, so as to obtain the target sequence corresponding to the target charge and discharge process.
Step 506, determining the battery RUL according to the set lifetime threshold and the target sequence of the target battery.
In the embodiment of the present disclosure, the set lifetime threshold may be used to represent a maximum value of the number of times the battery can be charged and discharged in a full life cycle, and the set lifetime threshold may be preset, for example, 10000, 15000, etc., which is not limited in the present disclosure.
It should be noted that, the set lifetime threshold may be determined based on an experimental basis, or may be determined according to actual needs, which is not limited in this disclosure.
In the embodiment of the present disclosure, the battery RUL may be determined according to the set lifetime threshold and the target order of the target battery.
It can be understood that the target sequence corresponding to the target charging and discharging process can represent the total number of times the target battery has been charged and discharged in the full life cycle. For example, the target charge and discharge process of the target battery G corresponds to a target order of 51, that is, the 51 th charge and discharge process in the whole life cycle of the target battery, which indicates that the total number of times the target battery G has been charged and discharged in the whole life cycle is 51.
As a possible implementation manner, the total number of times the target battery is charged and discharged may be determined according to a target sequence of target charging and discharging of the target battery; and determining the battery RUL of the target battery according to the difference between the set service life threshold value of the target battery and the total number of charged and discharged times.
As an example, assuming that the target order of target charge and discharge of the target battery is 51, the set lifetime threshold is 100, the total number of times the target battery has been charged and discharged is 51, and the battery RUL of the target battery is 49, that is, the remaining lifetime of the target battery is 49 charge and discharge processes, may be determined according to the difference 49 between the set lifetime threshold 100 and the total number of times the target battery has been charged and discharged.
According to the method for predicting the battery health SOH and the battery residual life RUL, a trained battery health SOH prediction model and a trained battery residual life RUL prediction model are obtained through a model training method adopting a battery SOH prediction model and an RUL prediction model; acquiring target operation data of a target charging and discharging process of a target battery; extracting features of the target operation data to obtain target key operation features of the target charging and discharging process; based on the target key operation characteristics, predicting the battery SOH of the target battery by adopting a trained battery SOH prediction model so as to obtain the target battery SOH; based on the SOH of the target battery, predicting the sequence of the target charging and discharging process by adopting a trained battery RUL prediction model so as to obtain a target sequence; and determining the battery RUL according to the set life threshold and the target sequence of the target battery. Therefore, the trained battery health degree SOH prediction model is adopted to identify the target key operation characteristics of the target charge and discharge process of the target battery so as to determine the target battery SOH of the target battery, and then the trained battery RUL prediction model is adopted to identify the target battery SOH so as to determine the target sequence of the target charge and discharge process, so that the battery RUL of the target battery can be effectively determined based on the target sequence.
Corresponding to the foregoing model training methods of the battery SOH prediction model and the RUL prediction model provided by the embodiments of fig. 1 to 4, the present disclosure further provides a model training apparatus of the battery SOH prediction model and the RUL prediction model, and since the model training apparatus of the battery SOH prediction model and the RUL prediction model provided by the embodiments of the present disclosure corresponds to the foregoing model training methods of the battery SOH prediction model and the RUL prediction model provided by the embodiments of fig. 1 to 4, the implementation of the model training methods of the battery SOH prediction model and the RUL prediction model are also applicable to the model training apparatus of the battery SOH prediction model and the RUL prediction model provided by the embodiments of the present disclosure, and will not be described in detail in the embodiments of the present disclosure.
Fig. 6 is a schematic structural diagram of a model training apparatus for a battery SOH prediction model and a RUL prediction model according to a fourth embodiment of the present disclosure.
As shown in fig. 6, the model training apparatus 600 of the battery SOH prediction model and the RUL prediction model may include: an acquisition module 601, an extraction module 602, a first training module 603 and a second training module 604.
The acquiring module 601 is configured to acquire historical operation data corresponding to a plurality of charging and discharging processes of the sample battery, and a corresponding first battery state of health SOH; any historical operation data comprises voltage information, current information and temperature information of the sample battery in the corresponding charging and discharging processes; the charging and discharging processes have a corresponding order.
The extracting module 602 is configured to perform feature extraction on the historical operation data corresponding to each charging and discharging process, so as to obtain key operation features of each charging and discharging process.
The first training module 603 is configured to perform model training on the battery SOH prediction model based on key operation features of each charging and discharging process and the corresponding first battery SOH, so as to obtain a trained battery SOH prediction model; the battery SOH prediction model is an extreme learning machine ELM model.
The second training module 604 is configured to perform model training on the battery RUL prediction model based on the sequence corresponding to each charging and discharging process and the first battery SOH, so as to obtain a trained battery RUL prediction model; the battery RUL prediction model is a model constructed based on a particle filter PF algorithm.
In one possible implementation of the embodiments of the present disclosure, the extracting module 602 is configured to: respectively extracting characteristics of historical operation data corresponding to each charging and discharging process to obtain initial operation characteristics of each charging and discharging process; the initial operation characteristics comprise voltage characteristics, current characteristics, temperature characteristics and electric quantity increment IC curve characteristics; and aiming at any charge and discharge process, carrying out feature analysis based on an extreme random tree algorithm on the initial operation features of the charge and discharge process so as to screen and retain the corresponding key operation features from the initial operation features of the charge and discharge process.
In one possible implementation of the embodiments of the present disclosure, the charging and discharging process includes a charging process, and the charging process includes a constant current charging stage and a constant voltage charging stage; an extraction module 602, configured to: aiming at any charge and discharge process, extracting characteristics of voltage information corresponding to a constant current charge stage in the charge and discharge process, and taking statistical characteristics of the extracted voltage information as voltage characteristics of the charge and discharge process; extracting characteristics of current information corresponding to a constant voltage charging stage in a charging and discharging process, and taking the statistical characteristics of the extracted current information as current characteristics in the charging and discharging process; extracting features of temperature information corresponding to a charging process in a charging and discharging process, so that statistical features of the extracted temperature information are used as temperature features of the charging and discharging process; generating a capacity increment IC curve according to current information and voltage information corresponding to a charging process in a charging and discharging process; and extracting the characteristics of the IC curve, and taking the statistical characteristics of the extracted IC curve as the characteristics of the IC curve in the charge-discharge process.
In one possible implementation of the disclosed embodiments, the statistical features of the voltage information include charge capacity; wherein, the charge capacity is obtained based on the PF algorithm.
In one possible implementation of the disclosed embodiments, a variable forgetting factor is employed in the extreme learning machine ELM model to adjust the output weights connecting each hidden layer node and each output layer node in the ELM model.
In one possible implementation of an embodiment of the present disclosure, a first training module 602 is configured to: based on key operation characteristics of each charging and discharging process and the corresponding first battery SOH, performing model training based on a whale optimization algorithm on the battery SOH prediction model to obtain a trained battery SOH prediction model.
In one possible implementation of an embodiment of the present disclosure, a first training module 602 is configured to: setting an initial value of a position value of each whale and a total number of whales of the whale population to generate a whale initial population; the position value of the whale represents the value of the super parameter of the battery SOH prediction model; performing an iteration process which is not more than a set iteration number based on a whale optimization algorithm on a battery SOH prediction model based on key operation characteristics of each charging and discharging process, a corresponding first battery SOH and a whale initial population, so as to update optimal whales in the whale population based on fitness values of each whale in any round of iteration process; wherein the fitness value of the optimal whale is not more than the fitness value of any whale except the optimal whale in the whale population; the fitness value of any whale is determined by inputting key operation characteristics of each charging and discharging process into a battery SOH prediction model adopting position values of corresponding whales and adopting an objective function according to the output of the battery SOH prediction model and the corresponding first battery SOH of each charging and discharging process.
In one possible implementation of an embodiment of the present disclosure, a first training module 602 is configured to:
aiming at any round of iterative process, under the condition that the sequence of the iterative process is not more than the set iterative times, acquiring the position value and the fitness value of the optimal whale of the previous round;
updating the position of any whale in the whale population by using the following formula, wherein the position value of the whale in the whale population after updating is used as the position value of the whale in the whale population in the present round:
；
；
wherein ,；
；
；
；
；
t is the sequence of the previous iteration process, and t+1 is the sequence of the present iteration process; the number of whales in the whale population is N; x is X woa (. Cndot.) is the position vector corresponding to the whale population of the corresponding iterative process, wherein X woa Any element in (-) is a position value corresponding to whale; x is X rand (t) is a randomly generated position vector corresponding to the whale population of the previous round, wherein X rand Any one of the elements (t) is a randomly generated position value for the corresponding whale;is a vector of size 1*N, and +.>The value of any element is equal to the position value of the optimal whale in the round; a is that woa(·) and Cwoa (. Cndot.) are all variable coefficients; d (D) woa (. Cndot.) represents the distance of movement; r is (r) woa (. Cndot.) is that the elements are all 0,1]A vector of random numbers in (a); b is a vector with elements of 1; t is t max To set the iteration number, b woa Is a constant equal to 1, l woa Is [0,1 ]]Random number, p woa (. Cndot.) is [0,1]N (0, 1) represents a gaussian function with an average value of 0 and a variance of 1;
determining an fitness value of the whale of the current round based on the position value of the whale for any whale of the whale population of the current round;
and determining the optimal whale of the round according to the fitness value of each whale of the round and the fitness value of the optimal whale of the previous round.
In one possible implementation of the embodiment of the present disclosure, the model training apparatus 600 of the battery SOH prediction model and the RUL prediction model may further include:
and the processing module is used for taking the position value of the optimal whale in the previous round as the optimal solution of the super parameter of the battery SOH prediction model and stopping the iterative process in the round under the condition that the sequence of the iterative process is larger than the set iterative times.
In one possible implementation of the embodiment of the disclosure, the second training module is configured to: based on the corresponding sequence of each charging and discharging process and the first battery SOH, model training based on a whale optimization algorithm is carried out on the battery RUL prediction model so as to obtain a trained battery RUL prediction model.
According to the model training device of the battery SOH prediction model and the RUL prediction model, historical operation data corresponding to a plurality of charging and discharging processes of a sample battery and corresponding first battery health state SOH are obtained; any historical operation data comprises voltage information, current information and temperature information of the sample battery in the corresponding charging and discharging processes; the charging and discharging processes have corresponding sequences; extracting features of historical operation data corresponding to each charging and discharging process respectively to obtain key operation features of each charging and discharging process; based on key operation characteristics of each charge and discharge process and the corresponding first battery SOH, performing model training on the battery SOH prediction model to obtain a trained battery SOH prediction model; the battery SOH prediction model is an extreme learning machine ELM model; model training is carried out on the battery RUL prediction model based on the corresponding sequence of each charging and discharging process and the first battery SOH so as to obtain a trained battery RUL prediction model; the battery RUL prediction model is a model constructed based on a particle filter PF algorithm. Therefore, the battery SOH prediction model and the battery RUL prediction model can be trained, the prediction effect of the model is improved, and the accuracy and reliability of the model prediction result are improved.
Corresponding to the method for predicting the battery state of health SOH and the battery remaining life RUL provided by the above-described fig. 5 embodiment, the present disclosure also provides an apparatus for predicting the battery state of health SOH and the battery remaining life RUL, and since the apparatus for predicting the battery state of health SOH and the battery remaining life RUL provided by the present disclosure embodiment corresponds to the method for predicting the battery state of health SOH and the battery remaining life RUL provided by the above-described fig. 5 embodiment, the implementation of the method for predicting the battery state of health SOH and the battery remaining life RUL is also applicable to the apparatus for predicting the battery state of health SOH and the battery remaining life RUL provided by the present disclosure embodiment, which will not be described in detail in the present disclosure embodiment.
Fig. 7 is a schematic structural diagram of an apparatus for predicting a battery state of health SOH and a remaining battery life RUL according to a fifth embodiment of the present disclosure.
As shown in fig. 7, the apparatus 700 for predicting the battery health SOH and the remaining battery life RUL may include: a first acquisition module 701, a second acquisition module 702, an extraction module 703, a first prediction module 704, a second prediction module 705, and a determination module 706.
The first obtaining module 701 is configured to obtain a trained battery health SOH prediction model and a trained battery remaining life RUL prediction model by using the model training method of the battery SOH prediction model and the RUL prediction model of any of the foregoing embodiments.
The second obtaining module 702 is configured to obtain target operation data of a target charging and discharging process of the target battery.
And the extracting module 703 is configured to perform feature extraction on the target operation data to obtain a target key operation feature of the target charging and discharging process.
The first prediction module 704 is configured to predict a battery SOH of the target battery based on the target key operation feature by using a trained battery SOH prediction model, so as to obtain the target battery SOH.
The second prediction module 705 is configured to predict, based on the target battery SOH, a target sequence of the target charge/discharge process using the trained battery RUL prediction model, so as to obtain the target sequence.
The determining module 706 is configured to determine the battery RUL according to the set lifetime threshold and the target order of the target battery.
According to the device for predicting the battery health SOH and the battery residual life RUL, a trained battery health SOH prediction model and a trained battery residual life RUL prediction model are obtained by adopting a model training method of the battery SOH prediction model and the RUL prediction model; acquiring target operation data of a target charging and discharging process of a target battery; extracting features of the target operation data to obtain target key operation features of the target charging and discharging process; based on the target key operation characteristics, predicting the battery SOH of the target battery by adopting a trained battery SOH prediction model so as to obtain the target battery SOH; based on the SOH of the target battery, predicting the sequence of the target charging and discharging process by adopting a trained battery RUL prediction model so as to obtain a target sequence; and determining the battery RUL according to the set life threshold and the target sequence of the target battery. Therefore, the trained battery health degree SOH prediction model is adopted to identify the target key operation characteristics of the target charge and discharge process of the target battery so as to determine the target battery SOH of the target battery, and then the trained battery RUL prediction model is adopted to identify the target battery SOH so as to determine the target sequence of the target charge and discharge process, so that the battery RUL of the target battery can be effectively determined based on the target sequence.
In order to achieve the above embodiment, the present invention further provides an electronic device, including: the system comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the model training method of the battery SOH prediction model and the RUL prediction model or the method for predicting the battery health SOH and the residual service life RUL of the battery according to any one of the embodiments.
In order to achieve the above-mentioned embodiments, the present invention also proposes a non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that the program, when executed by a processor, implements a model training method of a battery SOH prediction model and a RUL prediction model as proposed in any of the foregoing embodiments of the present invention, or a method for predicting a battery state of health SOH and a remaining life RUL of a battery.
In order to implement the above embodiments, the present invention also proposes a computer program product, which when executed by a processor, performs a model training method of a battery SOH prediction model and a RUL prediction model, or a method for predicting a battery state of health SOH and a remaining battery life RUL, as proposed by any of the previous embodiments of the present invention.
According to embodiments of the present invention, the present invention also provides an electronic device, a non-transitory computer-readable storage medium, and a computer program product.
As shown in fig. 8, the electronic device 12 is in the form of a general purpose computing device. Components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, a bus 18 that connects the various system components, including the system memory 28 and the processing units 16.
Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, and a local bus using any of a variety of bus architectures. By way of example, and not limitation, such architectures include industry Standard architecture (Industry Standard Architecture; hereinafter ISA) bus, micro channel architecture (Micro Channel Architecture; hereinafter MAC) bus, enhanced ISA bus, video electronics standards Association (Video Electronics Standards Association; hereinafter VESA) local bus, and peripheral component interconnect (Peripheral Component Interconnection; hereinafter PCI) bus.
Electronic device 12 typically includes a variety of computer system readable media. Such media can be any available media that is accessible by electronic device 12 and includes both volatile and nonvolatile media, removable and non-removable media.
Memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (Random Access Memory; hereinafter: RAM) 30 and/or cache memory 32. The electronic device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from or write to non-removable, nonvolatile magnetic media (not shown in FIG. 8, commonly referred to as a "hard disk drive"). Although not shown in fig. 8, a magnetic disk drive for reading from and writing to a removable non-volatile magnetic disk (e.g., a "floppy disk"), and an optical disk drive for reading from or writing to a removable non-volatile optical disk (e.g., a compact disk read only memory (Compact Disc Read Only Memory; hereinafter CD-ROM), digital versatile read only optical disk (Digital Video Disc Read Only Memory; hereinafter DVD-ROM), or other optical media) may be provided. In such cases, each drive may be coupled to bus 18 through one or more data medium interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to carry out the functions of embodiments of the invention.
A program/utility 40 having a set (at least one) of program modules 42 may be stored in, for example, memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each or some combination of which may include an implementation of a network environment. Program modules 42 generally perform the functions and/or methods of the embodiments described herein.
The electronic device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), one or more devices that enable a user to interact with the electronic device 12, and/or any devices (e.g., network card, modem, etc.) that enable the electronic device 12 to communicate with one or more other computing devices. Such communication may occur through an input/output (I/O) interface 22. Also, the electronic device 12 may communicate with one or more networks, such as a local area network (Local Area Network; hereinafter: LAN), a wide area network (Wide Area Network; hereinafter: WAN) and/or a public network, such as the Internet, via the network adapter 20. As shown, the network adapter 20 communicates with other modules of the electronic device 12 over the bus 18. It should be appreciated that although not shown, other hardware and/or software modules may be used in connection with electronic device 12, including, but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, data backup storage systems, and the like.
The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, for example, implementing the methods mentioned in the foregoing embodiments.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present invention, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and additional implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order from that shown or discussed, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present invention.
Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. As with the other embodiments, if implemented in hardware, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
Those of ordinary skill in the art will appreciate that all or a portion of the steps carried out in the method of the above-described embodiments may be implemented by a program to instruct related hardware, where the program may be stored in a computer readable storage medium, and where the program, when executed, includes one or a combination of the steps of the method embodiments.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing module, or each unit may exist alone physically, or two or more units may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a computer readable storage medium if implemented in the form of software functional modules and sold or used as a stand-alone product.
The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, or the like. While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention.
Claims (11)
1. A model training method for a battery SOH prediction model and a RUL prediction model, the method comprising:
acquiring historical operation data corresponding to a plurality of charging and discharging processes of a sample battery and corresponding first battery health states SOH; any one of the historical operation data comprises voltage information, current information and temperature information of the sample battery in the corresponding charging and discharging process; the charging and discharging processes have corresponding sequences;
extracting features of historical operation data corresponding to each charging and discharging process respectively to obtain key operation features of each charging and discharging process;
based on key operation characteristics of each charging and discharging process and the corresponding first battery SOH, performing model training on the battery SOH prediction model to obtain a trained battery SOH prediction model; the battery SOH prediction model is an extreme learning machine ELM model;
Model training is carried out on the battery RUL prediction model based on the sequence corresponding to the charging and discharging processes and the first battery SOH so as to obtain a trained battery RUL prediction model; the battery RUL prediction model is a model constructed based on a particle filter PF algorithm.
2. The method of claim 1, wherein the feature extraction is performed on the historical operating data corresponding to each of the charging and discharging processes to obtain key operating features of each of the charging and discharging processes, respectively, including:
respectively extracting characteristics of historical operation data corresponding to each charging and discharging process to obtain initial operation characteristics of each charging and discharging process; wherein the initial operating characteristics include voltage characteristics, current characteristics, temperature characteristics, and electrical delta IC curve characteristics;
and aiming at any charge and discharge process, carrying out feature analysis based on an extreme random tree algorithm on the initial operation features of the charge and discharge process so as to screen and retain the corresponding key operation features from the initial operation features of the charge and discharge process.
3. The method of claim 2, wherein the charge-discharge process comprises a charge process comprising a constant current charge phase and a constant voltage charge phase;
The feature extraction is performed on the historical operation data corresponding to each charging and discharging process to obtain the initial operation feature of each charging and discharging process, including:
performing feature extraction on voltage information corresponding to a constant current charging stage in the charging and discharging process aiming at any charging and discharging process, so as to take the statistical features of the extracted voltage information as the voltage features of the charging and discharging process;
extracting characteristics of current information corresponding to a constant voltage charging stage in the charging and discharging process, so that statistical characteristics of the extracted current information are used as current characteristics in the charging and discharging process;
feature extraction is carried out on temperature information corresponding to a charging process in the charging and discharging process, so that statistical features of the extracted temperature information are used as temperature features of the charging and discharging process;
generating a capacity increment IC curve according to current information and voltage information corresponding to a charging process in the charging and discharging processes;
and extracting the characteristics of the IC curve, and taking the statistical characteristics of the extracted IC curve as the characteristics of the IC curve in the charging and discharging process.
4. A method according to claim 3, wherein the statistical characteristics of the voltage information include charge capacity; wherein the charge capacity is obtained based on a PF algorithm.
5. The method of claim 1, wherein a variable forgetting factor is employed in the extreme learning machine ELM model to adjust output weights connecting hidden layer nodes and output layer nodes in the ELM model.
6. The method of claim 1, wherein model training the battery SOH prediction model based on the key operating characteristics of each of the charge-discharge processes and the corresponding first battery SOH to obtain a trained battery SOH prediction model comprises:
and carrying out model training based on a whale optimization algorithm on the battery SOH prediction model based on key operation characteristics of each charging and discharging process and the corresponding first battery SOH so as to obtain a trained battery SOH prediction model.
7. The method of claim 6, wherein the model training the battery SOH prediction model based on the whale optimization algorithm based on the key operating characteristics of each of the charge and discharge processes and the corresponding first battery SOH to obtain a trained battery SOH prediction model comprises:
setting an initial value of a position value of each whale and a total number of whales of the whale population to generate a whale initial population; the position value of the whale represents the value of the super parameter of the battery SOH prediction model;
Performing an iteration process based on a whale optimization algorithm and not more than a set iteration number on the battery SOH prediction model based on key operation characteristics of each charging and discharging process, a corresponding first battery SOH and the whale initial population, so as to update optimal whales in the whale population based on fitness values of each whale in any round of the iteration process; wherein the fitness value of the optimal whale is not greater than the fitness value of any of the whales in the whale population other than the optimal whale; the fitness value of any whale is determined by inputting key operation characteristics of each charging and discharging process into a battery SOH prediction model adopting position values of the corresponding whale and adopting an objective function according to the output of the battery SOH prediction model and the first battery SOH corresponding to each charging and discharging process.
8. The method of claim 7, wherein updating the optimal whale in the whale population based on the fitness value of each whale during any one of the rounds of the iterative process comprises:
aiming at any round of iterative process, under the condition that the sequence of the iterative process is not more than the set iterative times, acquiring the position value and the fitness value of the optimal whale of the previous round;
Updating the position of any one of the whales in the whale population by using the following formula, wherein the position value of the whale in the whale population after updating is used as the position value of the whale in the whale population in the current round:
；
；
wherein ,；
；
；
；
；
t is the sequence of the previous iteration process, and t+1 is the sequence of the present iteration process; the number of whales in the whale population is N; x is X woa (. Cndot.) is the position vector corresponding to the whale population of the corresponding iterative process, wherein X woa Any element in (-) is a position value corresponding to whale; x is X rand (t) isRandomly generated position vectors corresponding to the whale population of the previous round, wherein X rand Any one of the elements (t) is a randomly generated position value for the corresponding whale;is a vector of size 1*N, and +.>The value of any element is equal to the position value of the optimal whale in the round; a is that woa(·) and Cwoa (. Cndot.) are all variable coefficients; d (D) woa (. Cndot.) represents the distance of movement; r is (r) woa (. Cndot.) is that the elements are all 0,1]A vector of random numbers in (a); b is a vector with elements of 1; t is t max To set the iteration number, b woa Is a constant equal to 1, l woa Is [0,1 ]]Random number, p woa (. Cndot.) is [0,1]N (0, 1) represents a gaussian function with an average value of 0 and a variance of 1;
determining, for any whale of the whale population of the present round, an fitness value of the whale of the present round based on the position value of the whale;
And determining the optimal whale of the round according to the fitness value of each whale of the round and the fitness value of the optimal whale of the previous round.
9. The method of claim 7, wherein the method further comprises:
and under the condition that the sequence of the iterative process is larger than the set iterative times, taking the position value of the last round of optimal whale as the optimal solution of the super parameter of the battery SOH prediction model, and stopping the iterative process of the round.
10. The method of claim 1, wherein model training the battery RUL prediction model based on the sequence of each of the charge-discharge processes and the first battery SOH to obtain a trained battery RUL prediction model comprises:
and performing model training based on a whale optimization algorithm on the battery RUL prediction model based on the sequence corresponding to the charging and discharging processes and the first battery SOH so as to obtain a trained battery RUL prediction model.
11. A method for predicting battery SOH and battery RUL, the method comprising:
obtaining a trained battery state of health SOH prediction model and a battery remaining life RUL prediction model using the method of any one of claims 1-10;
Acquiring target operation data of a target charging and discharging process of a target battery;
extracting features of the target operation data to obtain target key operation features of the target charging and discharging process;
based on the target key operation characteristics, predicting the battery SOH of the target battery by adopting the trained battery SOH prediction model so as to obtain the target battery SOH;
based on the target battery SOH, predicting the sequence of the target charge-discharge process by adopting the trained battery RUL prediction model so as to obtain a target sequence;
and determining a target battery RUL of the target battery according to the set life threshold of the target battery and the target sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310871063.3A CN116593904B (en) | 2023-07-17 | 2023-07-17 | Model training method and method for predicting battery SOH and battery RUL |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310871063.3A CN116593904B (en) | 2023-07-17 | 2023-07-17 | Model training method and method for predicting battery SOH and battery RUL |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116593904A true CN116593904A (en) | 2023-08-15 |
| CN116593904B CN116593904B (en) | 2023-10-03 |
Family
ID=87608441
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310871063.3A Active CN116593904B (en) | 2023-07-17 | 2023-07-17 | Model training method and method for predicting battery SOH and battery RUL |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116593904B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103399276A (en) * | 2013-07-25 | 2013-11-20 | 哈尔滨工业大学 | Lithium-ion battery capacity estimation and residual cycling life prediction method |
| CN112098878A (en) * | 2020-09-15 | 2020-12-18 | 重庆邮电大学 | End-to-end neural network establishment method for lithium battery SOH estimation and RUL prediction |
| CN112834927A (en) * | 2021-01-06 | 2021-05-25 | 合肥工业大学 | Lithium battery residual life prediction method, system, device and medium |
| CN114814592A (en) * | 2022-03-14 | 2022-07-29 | 湖北工业大学 | Lithium battery health state estimation and remaining service life prediction method and equipment |
-
2023
- 2023-07-17 CN CN202310871063.3A patent/CN116593904B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103399276A (en) * | 2013-07-25 | 2013-11-20 | 哈尔滨工业大学 | Lithium-ion battery capacity estimation and residual cycling life prediction method |
| CN112098878A (en) * | 2020-09-15 | 2020-12-18 | 重庆邮电大学 | End-to-end neural network establishment method for lithium battery SOH estimation and RUL prediction |
| CN112834927A (en) * | 2021-01-06 | 2021-05-25 | 合肥工业大学 | Lithium battery residual life prediction method, system, device and medium |
| CN114814592A (en) * | 2022-03-14 | 2022-07-29 | 湖北工业大学 | Lithium battery health state estimation and remaining service life prediction method and equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116593904B (en) | 2023-10-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111443294B (en) | Method and device for indirectly predicting remaining life of lithium ion battery | |
| Li et al. | Lithium-ion battery capacity estimation—A pruned convolutional neural network approach assisted with transfer learning | |
| Luzi et al. | A novel neural networks ensemble approach for modeling electrochemical cells | |
| Takyi-Aninakwa et al. | An optimized relevant long short-term memory-squared gain extended Kalman filter for the state of charge estimation of lithium-ion batteries | |
| CN111812515A (en) | XGboost model-based lithium ion battery state of charge estimation | |
| CN114167284B (en) | Lithium battery RUL prediction method and equipment based on BMS big data and integrated learning | |
| CN112098873B (en) | Lithium battery health state estimation method based on charging voltage curve geometric characteristics | |
| Venugopal et al. | Analysis of optimal machine learning approach for battery life estimation of Li-ion cell | |
| CN115201686B (en) | Lithium ion battery health state assessment method under incomplete charge and discharge data | |
| CN114660497A (en) | Lithium ion battery service life prediction method aiming at capacity regeneration phenomenon | |
| Wu et al. | State of health estimation of lithium-ion batteries using Autoencoders and Ensemble Learning | |
| Savargaonkar et al. | A cycle-based recurrent neural network for state-of-charge estimation of li-ion battery cells | |
| CN114779089A (en) | Method for calculating battery state of charge based on energy storage lithium battery equivalent circuit model | |
| Xue et al. | A Flexible deep convolutional neural network coupled with progressive training framework for online capacity estimation of lithium-ion batteries | |
| CN112731183B (en) | Improved ELM-based lithium ion battery life prediction method | |
| CN116593904B (en) | Model training method and method for predicting battery SOH and battery RUL | |
| CN117310504A (en) | Battery health state evaluation method, system, equipment and medium | |
| CN116609673A (en) | Battery remaining life prediction method and device based on charging characteristics and transfer learning | |
| CN117074951A (en) | Lithium battery state of charge estimation method, device, equipment and readable storage medium | |
| CN116754949A (en) | Intelligent health state prediction method for lithium ion battery | |
| CN115856641A (en) | Method and device for predicting remaining charging time of battery and electronic equipment | |
| CN114740389A (en) | Battery health assessment method and device, electronic equipment and readable storage medium | |
| CN114966436A (en) | Lithium battery state of charge prediction method, device, equipment and readable storage medium | |
| CN114487890A (en) | Lithium battery health state estimation method for improving long-term and short-term memory neural network | |
| CN113743670B (en) | GRU model-based circuit fault real-time prediction method and verification circuit |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |