CN112798960A - Battery pack residual life prediction method based on migration deep learning - Google Patents
Battery pack residual life prediction method based on migration deep learning Download PDFInfo
- Publication number
- CN112798960A CN112798960A CN202110048627.4A CN202110048627A CN112798960A CN 112798960 A CN112798960 A CN 112798960A CN 202110048627 A CN202110048627 A CN 202110048627A CN 112798960 A CN112798960 A CN 112798960A
- Authority
- CN
- China
- Prior art keywords
- battery
- battery pack
- monomer
- model
- capacity
- 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
Images
Classifications
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
Abstract
The invention relates to a battery pack residual life prediction method based on migration deep learning, and belongs to the technical field of batteries. The method comprises the following steps: step S1: collecting a power battery aging data set, and establishing a battery aging database; step S2: extracting a plurality of health factors according to the aging data of the single battery, and screening the health factors according to correlation analysis and capacity estimation errors; step S3: training based on a battery monomer full life cycle aging data set to obtain a recursion model of a health factor and a capacity estimation model based on the health factor; step S4: establishing a machine learning model based on the battery monomer health factor set and the battery pack capacity attenuation; step S5: and predicting the capacity of each monomer in the future based on the monomer capacity estimation model to obtain the monomer capacity distribution of the battery pack circulating in the future. By combining transfer learning and deep learning, the existing complete information can be effectively utilized, and the residual life prediction precision of the battery pack is improved.
Description
Technical Field
The invention belongs to the technical field of batteries, and relates to a battery pack residual life prediction method based on migration deep learning.
Background
The remaining life prediction method of a battery can be generally classified into a model-based method and a data-driven method. One model-based approach is to make predictions primarily from empirical or semi-empirical models built from historical data and cycle times. Typically an exponential model, a bi-exponential model, or a polynomial model. And performing curve fitting by using an advanced filter such as Kalman filtering, particle filtering and the like to obtain a fitting curve, and thus performing capacity estimation or residual life prediction by using the fitting curve. The method is another model-based method, and an aging mechanism model of the battery is established, so that the charge and discharge simulation of future circulation is realized through simulation, and further the residual life when the capacity is attenuated to a threshold value is obtained. Data-driven based methods have been rapidly developed in recent years because they do not require special models, but rely only on the characteristics of the data itself. Data-driven based methods typically construct a mapping with a sequence of capacity fade, predict the next capacity data from the previous ones, and extrapolate to the remaining cycle life. Or extracting related health factors according to special working conditions in the charging and discharging processes, establishing a mapping relation between the health factors and the capacity, and estimating the capacity of the battery through the health factors. And then establishing a battery capacity sequence mapping relation and extrapolating to obtain the residual cycle life of the battery. However, current research is still lacking in the realization of life prediction of a battery pack. The migration learning can utilize effective historical information to carry out small correction on the model of the prediction task, so that the effect of the prediction task is improved. At present, research still lacks of effectively utilizing the corresponding relation between a single battery cell and a battery pack, so that the prediction precision of the residual service life of the battery pack is improved. In addition, the remaining life of the battery pack not only needs to pay attention to future capacity changes of the whole pack, but also needs to pay attention to capacity distribution of each battery cell, so that the battery cells with high inconsistency can be identified, and timely replacement or balanced management can be performed, so that the service life of the battery pack is prolonged.
In order to solve the problems, an effective efficient and accurate prediction method for the remaining life of the battery pack and the service life distribution of the battery pack monomer based on the migration deep learning is not provided at present.
Disclosure of Invention
In view of the above, the present invention provides a method for predicting remaining life of a battery pack based on deep migration learning.
In order to achieve the purpose, the invention provides the following technical scheme:
a battery pack residual life prediction method based on migration deep learning comprises the following steps:
step S1: collecting a power battery aging data set, and establishing a battery aging database which comprises complete battery monomer full-life cycle data and early aging data of a battery pack consisting of the same type of battery cells;
step S2: extracting a plurality of health factors according to the aging data of the single battery, and screening the health factors according to correlation analysis and capacity estimation errors;
step S3: training based on a battery monomer full life cycle aging data set to obtain a recursion model of a health factor and a capacity estimation model based on the health factor;
step S4: extracting the health factor of each battery monomer in the battery pack early aging experimental data set, and establishing a machine learning model based on the health factor set of the battery monomers and the battery pack capacity attenuation;
step S5: carrying out fine adjustment on a health factor recursion model of each battery monomer of the battery pack by utilizing transfer learning, carrying out health factor extrapolation prediction, and finally estimating the capacity of the battery pack in future circulation based on a monomer health factor set obtained by extrapolation; and predicting the capacity of each monomer in the future based on the monomer capacity estimation model to obtain the monomer capacity distribution of the battery pack circulating in the future.
Optionally, step S1 specifically includes:
step S11: collecting a complete data set of a whole life cycle of a plurality of battery monomers of a certain type of battery cell, wherein the data set comprises parameters of charge and discharge current, voltage, temperature, time and electric quantity, and covers data sets of different charge and discharge multiplying powers and environmental temperatures;
step S12: collecting an aging experiment data set of a battery pack corresponding to the same type of battery cell of each battery monomer, wherein the aging experiment data set comprises voltage and temperature parameters of each monomer, and current and electric quantity information of the battery pack;
step S13: and establishing a battery aging data set of a certain type of battery cell according to the collected aging experiment data of the battery monomer and the battery pack.
Optionally, step S2 specifically includes:
step S21: extracting a plurality of available health factors according to the whole life cycle of the battery monomer;
step S22: screening out health factors with high correlation with the capacity of the battery monomer through correlation analysis;
step S23: a machine learning model of the health factor estimated capacity of the whole life is established based on the battery monomer A, the health factor is extracted by the battery monomer B for capacity estimation, and estimation errors are analyzed.
Optionally, the plurality of available health factors include a slope of a voltage curve, a voltage/temperature change at equal time intervals, a power change at equal voltage/temperature intervals, a power sequence variance, a power difference variance, a peak value, a valley value, a peak interval, a peak area, a peak value, a valley value, and a peak interval of a voltage/temperature difference curve;
the battery monomer A and the battery monomer B refer to different battery monomers of the same type of battery core, and the machine learning model is a Gaussian process regression model or a correlation vector machine model, so that probability prediction is realized.
Optionally, step S3 specifically includes:
step S31: establishing a deep learning model of the recursive attenuation of the life cycle health factors of the plurality of monomers according to the selected health factors;
step S32: and establishing a health factor-based capacity estimation model of the whole life cycle of the plurality of monomers according to the selected health factor.
Optionally, the deep learning model specifically refers to a model constructed by a multilayer neural network, and is a gaussian process regression model or a correlation vector machine model, so as to implement probability prediction.
Optionally, step S4 specifically includes:
step S41: extracting the health factor of each battery monomer in the early cycle of the battery pack according to the selected health factor, and establishing a feature set formed by the health factors of each monomer of the battery pack;
step S42: and establishing a machine learning model for estimating the capacity of the battery pack according to the monomer feature set and the true value of the capacity of the battery pack.
Optionally, the feature set is a feature matrix formed by features of each battery cell in the battery pack, and is a gaussian process regression model or a correlation vector machine model, so as to implement probability prediction.
Optionally, step S5 specifically includes:
step S51: the early-stage trained life cycle health factor attenuation model of the battery monomer and the health factors extracted from each monomer of the battery pack at the early stage are utilized, and a transfer learning method is used for retraining part of the structure of the trained network to obtain a health factor recursion model of each monomer of the battery pack;
step S52: extrapolating to obtain a health factor predicted value of future unknown cycle times according to the health factor recurrence model of each monomer;
step S53: estimating a machine learning model of the battery pack capacity according to the established battery monomer characteristic set, and predicting the capacity value of the battery pack circulating in the future;
step S54: according to the established battery monomer capacity estimation model, obtaining a capacity prediction value of each monomer in the battery pack with unknown cycle in the future so as to obtain the capacity prediction distribution of the battery pack monomers;
the method for the transfer learning selectively freezes certain layers based on a deep learning network of an early-trained health factor recursion model, and retrains the model of the rest layers, or removes certain layers and reconstructs the network again to retrain a new network.
The invention has the beneficial effects that:
1) by combining transfer learning and deep learning, the existing complete information can be effectively utilized, and the residual life prediction precision of the battery pack is improved.
2) The provided residual life prediction of the battery pack can realize the prediction of the whole pack life and the prediction of the life distribution of the single batteries.
3) The proposed solution can guide the design and development of battery packs, using only early data for life prediction, thereby optimizing the design of the battery packs.
4) The proposed solution provides probabilistic predictions, providing a confidence distribution of the prediction.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention may be realized and attained by the means of the instrumentalities and combinations particularly pointed out hereinafter.
Drawings
For the purposes of promoting a better understanding of the objects, aspects and advantages of the invention, reference will now be made to the following detailed description taken in conjunction with the accompanying drawings in which:
FIG. 1 is a process flow diagram of the present invention as a whole;
FIG. 2 is a schematic diagram of a health factor of an embodiment; (a) extracting a schematic diagram based on the measured data characteristics; (b) is a characteristic diagram based on the variance of electric quantity; (c) extracting a schematic diagram based on IC curve characteristics; (d) extracting a schematic diagram based on DV curve characteristics; (e) extracting a schematic diagram based on DT curve characteristics;
FIG. 3 is a technical roadmap of an embodiment;
FIG. 4 is a diagram of a migration deep learning network architecture employed in an embodiment;
FIG. 5 is a flow chart of health factor screening and model building according to an embodiment.
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It should be noted that the drawings provided in the following embodiments are only for illustrating the basic idea of the present invention in a schematic way, and the features in the following embodiments and examples may be combined with each other without conflict.
Wherein the showings are for the purpose of illustrating the invention only and not for the purpose of limiting the same, and in which there is shown by way of illustration only and not in the drawings in which there is no intention to limit the invention thereto; to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product; it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The same or similar reference numerals in the drawings of the embodiments of the present invention correspond to the same or similar components; in the description of the present invention, it should be understood that if there is an orientation or positional relationship indicated by terms such as "upper", "lower", "left", "right", "front", "rear", etc., based on the orientation or positional relationship shown in the drawings, it is only for convenience of description and simplification of description, but it is not an indication or suggestion that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes, and are not to be construed as limiting the present invention, and the specific meaning of the terms may be understood by those skilled in the art according to specific situations.
Referring to fig. 1, a method for predicting remaining life of a battery pack based on deep migration learning may include the following steps:
step S1: and collecting a power battery aging data set, and establishing a battery aging database which comprises complete battery monomer full-life cycle data and early aging data of a battery pack consisting of the same type of battery cells.
Step S2: and extracting a plurality of health factors according to the aging data of the single batteries, and screening the health factors according to correlation analysis and capacity estimation errors.
Step S3: training based on the battery monomer full life cycle aging data set to obtain a recursion model of the health factor and a capacity estimation model based on the health factor.
Step S4: and extracting the health factor of each battery cell in the early aging experimental data set of the battery pack, and establishing a machine learning model based on the health factor set of the battery cells and the capacity attenuation of the battery pack.
Step S5: carrying out fine adjustment on a health factor recursion model of each battery monomer of the battery pack by utilizing transfer learning, carrying out health factor extrapolation prediction, and finally estimating the capacity of the battery pack in future circulation based on a monomer health factor set obtained by extrapolation; and predicting the capacity of each monomer in the future based on the monomer capacity estimation model to obtain the monomer capacity distribution of the battery pack circulating in the future.
As an alternative embodiment, the complete technical route diagram of the present solution is shown in fig. 3.
As an alternative embodiment, the step S1 specifically includes steps S11-S13:
step S11: the method comprises the steps of collecting complete data sets of the whole life cycles of a plurality of battery monomers of a certain type of battery cell, wherein the data sets comprise parameters such as charging and discharging current, voltage, temperature, time, electric quantity and the like, and can cover data sets of different charging and discharging multiplying powers and environmental temperatures.
Step S12: and collecting an aging experiment data set of the battery pack corresponding to the battery monomer and the same type of battery core, wherein the aging experiment data set comprises parameters such as voltage and temperature of each monomer and information such as current and electric quantity of the battery pack.
Step S13: and establishing a battery aging data set of a certain type of battery cell according to the collected aging experiment data of the battery monomer and the battery pack.
As an alternative embodiment, the step S2 specifically includes steps S21-S23:
step S21: and extracting a plurality of available health factors according to the whole life cycle of the battery cell.
Step S22: and screening out health factors with high correlation with the capacity of the battery cell through correlation analysis.
Step S23: a machine learning model of the health factor estimated capacity of the whole life is established based on the battery monomer A, the health factor is extracted by the battery monomer B for capacity estimation, and estimation errors are analyzed.
As an alternative embodiment, the plurality of available health factors in S21 include a slope of a voltage curve, a voltage/temperature variation at equal time intervals, a power variation at equal voltage/temperature intervals, a power sequence variance, a power difference variance, a peak value, a valley value, a peak interval, a peak area of a capacity increment curve, a peak value, a valley value, a peak interval, and the like, as shown in fig. 2, which are schematic diagrams of health factors of the embodiment; (a) extracting a schematic diagram based on the measured data characteristics; (b) is a characteristic diagram based on the variance of electric quantity; (c) extracting a schematic diagram based on IC curve characteristics; (d) extracting a schematic diagram based on DV curve characteristics; (e) extracting a schematic diagram based on DT curve characteristics;
the extraction method comprises the following steps:
the slope of the voltage curve;
b) same charge/discharge time voltage difference: ETDV ═ f (V)0,t_interval)
c) The electric quantity difference of the same voltage interval; EVDQ ═ f (Q)0,t_interval)
d) Different circulating electric quantity difference variances;
ΔQ＝Qci-Qcj、std_ΔQ＝std(Qci-Qcj)
e) incremental Capacity (IC) curve characteristics (peak, valley, peak-to-peak distance, peak-to-peak area)Product, etc.); IC incremental calculation:
f) differential Voltage (DV) curve characteristics (valley, peak-to-peak, etc.); differential voltage calculation formula:
g) differential Temperature (DT) curve characteristics (valley, peak-to-peak, etc.). Differential temperature calculation:
as an optional embodiment, the process of obtaining the IC, DV, and DT curves includes filtering and denoising, and gaussian filtering is adopted:
as an alternative example, the correlation coefficient method in the evaluation system described in S22 may use a pearson correlation coefficient, as follows
Is represented by the formula:
as an alternative embodiment, a Gaussian process regression is adopted to model by using a battery No. A and estimate by using a battery No. B
The effect is that the Gaussian process regression algorithm flow is as follows:
the input and output are assumed to conform to the following bayesian multivariate regression model: y ═ f (x) + epsilon,
where ε is white noise that fits a Gaussian distribution. (x) can be written as:
where m (x) and k (x, x') are the mean function and covariance function, respectively:
m(x)＝E[f(x)]，k(x,x')＝E[(f(x)-m(x))(f(x')-m(x'))T]
the input-output relationship can be written as:
in the formula InFor an n-dimensional identity matrix, the output mean and error covariance of the GPR can be written as:
as an alternative embodiment, the estimation effect is evaluated using the root mean square error.
As an alternative embodiment, the step S3 specifically includes steps S31-S32
Step S31: and establishing a deep learning model of the recursive attenuation of the life cycle health factors of the plurality of monomers according to the selected health factors.
Step S32: and establishing a health factor-based capacity estimation model of the whole life cycle of the plurality of monomers according to the selected health factor.
As an alternative embodiment, the life cycle health factor recurrence decay model adopts the way that the first m health factors predict the health factor of the next cycle, namely: HI (high-intensity)k+1＝f(HIk-m,...,HIk) The life cycle health factor recursive attenuation model is modeled by using a deep neural network, and specifically comprises an input layer, a long-time memory neural network layer, a full connection layer and an output layer as shown in fig. 4. The capacity estimation model also adopts a Gaussian process regression model.
As an alternative embodiment, step S4 specifically includes steps S41-S42
Step S41: and extracting the health factor of each battery monomer in the early circulation of the battery pack according to the selected health factor, and establishing a feature set consisting of the health factors of each battery monomer.
Step S42: and establishing a machine learning model for estimating the capacity of the battery pack according to the monomer feature set and the true value of the capacity of the battery pack.
As an optional embodiment, the feature set is a feature matrix established by taking the health factors of each battery cell after being screened as one column and the cycle number as a row, and a battery pack capacity estimation model based on the feature set is established by adopting gaussian process regression.
The specific technical implementation of steps S2-S4 is shown in FIG. 5.
As an alternative embodiment, step S5 specifically includes steps S51-S54
Step S51: and (3) performing partial structure retraining on the trained network by using the early-trained life cycle health factor attenuation model of the battery monomer and the health factors extracted at the early stage of each monomer of the battery pack by using a transfer learning method to obtain a health factor recurrence model of each monomer of the battery pack.
Step S52: and (4) extrapolating to obtain a health factor predicted value of future unknown cycle times according to the health factor recurrence model of each monomer.
Step S53: and estimating a machine learning model of the battery pack capacity according to the established battery monomer characteristic set, and predicting the battery pack capacity value of the future cycle.
Step S54: and obtaining the predicted value of the capacity of each monomer in the battery pack of unknown cycle in the future according to the established battery monomer capacity estimation model, thereby obtaining the capacity prediction distribution of the battery pack monomers.
As an optional embodiment, the migration learning freezes the front input layer and the long-term memory neural network layer, and the parameters of the full connection layer and the output layer are retrained when each monomer is retrained, so that the purpose of model fine tuning is achieved.
Finally, the above embodiments are only intended to illustrate the technical solutions of the present invention and not to limit the present invention, and although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions, and all of them should be covered by the claims of the present invention.
Claims (9)
1. A battery pack residual life prediction method based on migration deep learning is characterized in that: the method comprises the following steps:
step S1: collecting a power battery aging data set, and establishing a battery aging database which comprises complete battery monomer full-life cycle data and early aging data of a battery pack consisting of the same type of battery cells;
step S2: extracting a plurality of health factors according to the aging data of the single battery, and screening the health factors according to correlation analysis and capacity estimation errors;
step S3: training based on a battery monomer full life cycle aging data set to obtain a recursion model of a health factor and a capacity estimation model based on the health factor;
step S4: extracting the health factor of each battery monomer in the battery pack early aging experimental data set, and establishing a machine learning model based on the health factor set of the battery monomers and the battery pack capacity attenuation;
step S5: carrying out fine adjustment on a health factor recursion model of each battery monomer of the battery pack by utilizing transfer learning, carrying out health factor extrapolation prediction, and finally estimating the capacity of the battery pack in future circulation based on a monomer health factor set obtained by extrapolation; and predicting the capacity of each monomer in the future based on the monomer capacity estimation model to obtain the monomer capacity distribution of the battery pack circulating in the future.
2. The battery pack remaining life prediction method based on the migration deep learning according to claim 1, characterized in that: the step S1 specifically includes:
step S11: collecting a complete data set of a whole life cycle of a plurality of battery monomers of a certain type of battery cell, wherein the data set comprises parameters of charge and discharge current, voltage, temperature, time and electric quantity, and covers data sets of different charge and discharge multiplying powers and environmental temperatures;
step S12: collecting an aging experiment data set of a battery pack corresponding to the same type of battery cell of each battery monomer, wherein the aging experiment data set comprises voltage and temperature parameters of each monomer, and current and electric quantity information of the battery pack;
step S13: and establishing a battery aging data set of a certain type of battery cell according to the collected aging experiment data of the battery monomer and the battery pack.
3. The battery pack remaining life prediction method based on the migration deep learning according to claim 2, characterized in that: the step S2 specifically includes:
step S21: extracting a plurality of available health factors according to the whole life cycle of the battery monomer;
step S22: screening out health factors with high correlation with the capacity of the battery monomer through correlation analysis;
step S23: a machine learning model of the health factor estimated capacity of the whole life is established based on the battery monomer A, the health factor is extracted by the battery monomer B for capacity estimation, and estimation errors are analyzed.
4. The battery pack remaining life prediction method based on the migration deep learning according to claim 3, characterized in that: the plurality of available health factors include a voltage curve slope, an equal time interval voltage/temperature change, an equal voltage/temperature interval power change, a power sequence variance, a power difference variance, a capacity increment curve peak value, a valley value, a peak interval, a peak area, a voltage/temperature difference curve peak value, a valley value and a peak interval;
the battery monomer A and the battery monomer B refer to different battery monomers of the same type of battery core, and the machine learning model is a Gaussian process regression model or a correlation vector machine model, so that probability prediction is realized.
5. The battery pack remaining life prediction method based on the migration deep learning according to claim 3, characterized in that: the step S3 specifically includes:
step S31: establishing a deep learning model of the recursive attenuation of the life cycle health factors of the plurality of monomers according to the selected health factors;
step S32: and establishing a health factor-based capacity estimation model of the whole life cycle of the plurality of monomers according to the selected health factor.
6. The battery pack remaining life prediction method based on the migration deep learning according to claim 5, characterized in that: the deep learning model is a model constructed by a multilayer neural network, is a Gaussian process regression model or a correlation vector machine model, and realizes probability prediction.
7. The battery pack remaining life prediction method based on the migration deep learning according to claim 5, characterized in that: the step S4 specifically includes:
step S41: extracting the health factor of each battery monomer in the early cycle of the battery pack according to the selected health factor, and establishing a feature set formed by the health factors of each monomer of the battery pack;
step S42: and establishing a machine learning model for estimating the capacity of the battery pack according to the monomer feature set and the true value of the capacity of the battery pack.
8. The battery pack remaining life prediction method based on the migration deep learning according to claim 7, characterized in that: the characteristic set is a characteristic matrix formed by the characteristics of each battery monomer in the battery pack and is a Gaussian process regression model or a correlation vector machine model, and probability prediction is achieved.
9. The battery pack remaining life prediction method based on the migration deep learning according to claim 7, characterized in that: the step S5 specifically includes:
step S51: the early-stage trained life cycle health factor attenuation model of the battery monomer and the health factors extracted from each monomer of the battery pack at the early stage are utilized, and a transfer learning method is used for retraining part of the structure of the trained network to obtain a health factor recursion model of each monomer of the battery pack;
step S52: extrapolating to obtain a health factor predicted value of future unknown cycle times according to the health factor recurrence model of each monomer;
step S53: estimating a machine learning model of the battery pack capacity according to the established battery monomer characteristic set, and predicting the capacity value of the battery pack circulating in the future;
step S54: according to the established battery monomer capacity estimation model, obtaining a capacity prediction value of each monomer in the battery pack with unknown cycle in the future so as to obtain the capacity prediction distribution of the battery pack monomers;
the method for the transfer learning selectively freezes certain layers based on a deep learning network of an early-trained health factor recursion model, and retrains the model of the rest layers, or removes certain layers and reconstructs the network again to retrain a new network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110048627.4A CN112798960B (en) | 2021-01-14 | 2021-01-14 | Battery pack residual life prediction method based on migration deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110048627.4A CN112798960B (en) | 2021-01-14 | 2021-01-14 | Battery pack residual life prediction method based on migration deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112798960A true CN112798960A (en) | 2021-05-14 |
| CN112798960B CN112798960B (en) | 2022-06-24 |
Family
ID=75810843
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110048627.4A Active CN112798960B (en) | 2021-01-14 | 2021-01-14 | Battery pack residual life prediction method based on migration deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112798960B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113406496A (en) * | 2021-05-26 | 2021-09-17 | 广州市香港科大霍英东研究院 | Battery capacity prediction method, system, device and medium based on model migration |
| CN113791351A (en) * | 2021-09-17 | 2021-12-14 | 电子科技大学 | Lithium battery life prediction method based on transfer learning and difference probability distribution |
| CN113820604A (en) * | 2021-08-30 | 2021-12-21 | 昆明理工大学 | Lithium battery SOH estimation method based on temperature prediction |
| CN114167284A (en) * | 2021-11-02 | 2022-03-11 | 江苏博强新能源科技股份有限公司 | Lithium battery RUL prediction method and device based on BMS big data and ensemble learning |
| CN114184972A (en) * | 2021-11-02 | 2022-03-15 | 江苏博强新能源科技股份有限公司 | Method and equipment for automatically estimating SOH (state of health) of battery by combining data driving with electrochemical mechanism |
| CN114216558A (en) * | 2022-02-24 | 2022-03-22 | 西安因联信息科技有限公司 | Method and system for predicting remaining life of battery of wireless vibration sensor |
| CN116593903A (en) * | 2023-07-17 | 2023-08-15 | 中国华能集团清洁能源技术研究院有限公司 | Battery remaining life prediction method and device |
| WO2023189368A1 (en) * | 2022-03-30 | 2023-10-05 | ヌヴォトンテクノロジージャパン株式会社 | Storage battery degradation estimation device and storage battery degradation estimation method |
| CN113820604B (en) * | 2021-08-30 | 2024-04-26 | 昆明理工大学 | Lithium battery SOH estimation method based on temperature prediction |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150349385A1 (en) * | 2014-04-01 | 2015-12-03 | Medtronic, Inc. | Method and System for Predicting Useful Life of a Rechargeable Battery |
| CN107064800A (en) * | 2016-11-29 | 2017-08-18 | 北京交通大学 | The real-time predicting method of lithium ion battery remaining life |
| CN107797067A (en) * | 2016-09-05 | 2018-03-13 | 北京航空航天大学 | Lithium ion battery life migration prediction method based on deep learning |
| CN109061504A (en) * | 2018-08-28 | 2018-12-21 | 中北大学 | Same type difference lithium ion battery remaining life prediction technique and system |
| CN109342949A (en) * | 2018-11-06 | 2019-02-15 | 长沙理工大学 | Lithium-ion-power cell remaining life on-line prediction method in charging process |
| US20190257886A1 (en) * | 2018-02-21 | 2019-08-22 | Nec Laboratories America, Inc. | Deep learning approach for battery aging model |
| CN110927591A (en) * | 2019-12-11 | 2020-03-27 | 北京理工大学 | Battery capacity estimation method, computer readable medium and vehicle |
| CN111007417A (en) * | 2019-12-06 | 2020-04-14 | 重庆大学 | Battery pack SOH and RUL prediction method and system based on inconsistency evaluation |
| CN111443294A (en) * | 2020-04-10 | 2020-07-24 | 华东理工大学 | Method and device for indirectly predicting remaining life of lithium ion battery |
| CN111638465A (en) * | 2020-05-29 | 2020-09-08 | 浙大宁波理工学院 | Lithium battery health state estimation method based on convolutional neural network and transfer learning |
| CN111722115A (en) * | 2019-03-18 | 2020-09-29 | 上海汽车集团股份有限公司 | Power battery service life prediction method and system |
| CN111983474A (en) * | 2020-08-25 | 2020-11-24 | 陕西科技大学 | Lithium ion battery life prediction method and system based on capacity decline model |
| CN112014735A (en) * | 2019-05-30 | 2020-12-01 | 上海汽车集团股份有限公司 | Battery cell aging life prediction method and device based on full life cycle |
| CN112051506A (en) * | 2020-08-28 | 2020-12-08 | 北京航空航天大学 | Similar product transferable sample screening method, system and application |
| CN112083337A (en) * | 2020-10-22 | 2020-12-15 | 重庆大学 | Power battery health prediction method oriented to predictive operation and maintenance |
-
2021
- 2021-01-14 CN CN202110048627.4A patent/CN112798960B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150349385A1 (en) * | 2014-04-01 | 2015-12-03 | Medtronic, Inc. | Method and System for Predicting Useful Life of a Rechargeable Battery |
| CN107797067A (en) * | 2016-09-05 | 2018-03-13 | 北京航空航天大学 | Lithium ion battery life migration prediction method based on deep learning |
| CN107064800A (en) * | 2016-11-29 | 2017-08-18 | 北京交通大学 | The real-time predicting method of lithium ion battery remaining life |
| US20190257886A1 (en) * | 2018-02-21 | 2019-08-22 | Nec Laboratories America, Inc. | Deep learning approach for battery aging model |
| CN109061504A (en) * | 2018-08-28 | 2018-12-21 | 中北大学 | Same type difference lithium ion battery remaining life prediction technique and system |
| CN109342949A (en) * | 2018-11-06 | 2019-02-15 | 长沙理工大学 | Lithium-ion-power cell remaining life on-line prediction method in charging process |
| CN111722115A (en) * | 2019-03-18 | 2020-09-29 | 上海汽车集团股份有限公司 | Power battery service life prediction method and system |
| CN112014735A (en) * | 2019-05-30 | 2020-12-01 | 上海汽车集团股份有限公司 | Battery cell aging life prediction method and device based on full life cycle |
| CN111007417A (en) * | 2019-12-06 | 2020-04-14 | 重庆大学 | Battery pack SOH and RUL prediction method and system based on inconsistency evaluation |
| CN110927591A (en) * | 2019-12-11 | 2020-03-27 | 北京理工大学 | Battery capacity estimation method, computer readable medium and vehicle |
| CN111443294A (en) * | 2020-04-10 | 2020-07-24 | 华东理工大学 | Method and device for indirectly predicting remaining life of lithium ion battery |
| CN111638465A (en) * | 2020-05-29 | 2020-09-08 | 浙大宁波理工学院 | Lithium battery health state estimation method based on convolutional neural network and transfer learning |
| CN111983474A (en) * | 2020-08-25 | 2020-11-24 | 陕西科技大学 | Lithium ion battery life prediction method and system based on capacity decline model |
| CN112051506A (en) * | 2020-08-28 | 2020-12-08 | 北京航空航天大学 | Similar product transferable sample screening method, system and application |
| CN112083337A (en) * | 2020-10-22 | 2020-12-15 | 重庆大学 | Power battery health prediction method oriented to predictive operation and maintenance |
Non-Patent Citations (1)
| Title |
|---|
| 刘大同 等: "锂离子电池组健康状态估计综述", 《仪器仪表学报》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113406496A (en) * | 2021-05-26 | 2021-09-17 | 广州市香港科大霍英东研究院 | Battery capacity prediction method, system, device and medium based on model migration |
| CN113406496B (en) * | 2021-05-26 | 2023-02-28 | 广州市香港科大霍英东研究院 | Battery capacity prediction method, system, device and medium based on model migration |
| CN113820604A (en) * | 2021-08-30 | 2021-12-21 | 昆明理工大学 | Lithium battery SOH estimation method based on temperature prediction |
| CN113820604B (en) * | 2021-08-30 | 2024-04-26 | 昆明理工大学 | Lithium battery SOH estimation method based on temperature prediction |
| CN113791351B (en) * | 2021-09-17 | 2022-04-19 | 电子科技大学 | Lithium battery life prediction method based on transfer learning and difference probability distribution |
| CN113791351A (en) * | 2021-09-17 | 2021-12-14 | 电子科技大学 | Lithium battery life prediction method based on transfer learning and difference probability distribution |
| CN114167284A (en) * | 2021-11-02 | 2022-03-11 | 江苏博强新能源科技股份有限公司 | Lithium battery RUL prediction method and device based on BMS big data and ensemble learning |
| CN114184972B (en) * | 2021-11-02 | 2023-12-22 | 江苏博强新能源科技股份有限公司 | Automatic estimation method and equipment for SOH of battery by combining data driving and electrochemical mechanism |
| CN114167284B (en) * | 2021-11-02 | 2023-12-22 | 江苏博强新能源科技股份有限公司 | Lithium battery RUL prediction method and equipment based on BMS big data and integrated learning |
| CN114184972A (en) * | 2021-11-02 | 2022-03-15 | 江苏博强新能源科技股份有限公司 | Method and equipment for automatically estimating SOH (state of health) of battery by combining data driving with electrochemical mechanism |
| CN114216558B (en) * | 2022-02-24 | 2022-06-14 | 西安因联信息科技有限公司 | Method and system for predicting remaining life of battery of wireless vibration sensor |
| CN114216558A (en) * | 2022-02-24 | 2022-03-22 | 西安因联信息科技有限公司 | Method and system for predicting remaining life of battery of wireless vibration sensor |
| WO2023189368A1 (en) * | 2022-03-30 | 2023-10-05 | ヌヴォトンテクノロジージャパン株式会社 | Storage battery degradation estimation device and storage battery degradation estimation method |
| CN116593903A (en) * | 2023-07-17 | 2023-08-15 | 中国华能集团清洁能源技术研究院有限公司 | Battery remaining life prediction method and device |
| CN116593903B (en) * | 2023-07-17 | 2023-10-20 | 中国华能集团清洁能源技术研究院有限公司 | Battery remaining life prediction method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112798960B (en) | 2022-06-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112798960B (en) | Battery pack residual life prediction method based on migration deep learning | |
| CN111443294B (en) | Method and device for indirectly predicting remaining life of lithium ion battery | |
| Sui et al. | A review of non-probabilistic machine learning-based state of health estimation techniques for Lithium-ion battery | |
| CN110221225B (en) | Spacecraft lithium ion battery cycle life prediction method | |
| CN108896914B (en) | Gradient lifting tree modeling and prediction method for health condition of lithium battery | |
| CN110568359B (en) | Lithium battery residual life prediction method | |
| Stephenson et al. | Bayesian inference for extremes: accounting for the three extremal types | |
| CN111680848A (en) | Battery life prediction method based on prediction model fusion and storage medium | |
| CN110658462B (en) | Lithium battery online service life prediction method based on data fusion and ARIMA model | |
| CN103954913A (en) | Predication method of electric vehicle power battery service life | |
| CN111458646A (en) | Lithium battery SOC estimation method based on PSO-RBF neural network | |
| CN112611976A (en) | Power battery state of health estimation method based on double differential curves | |
| Venugopal et al. | Analysis of optimal machine learning approach for battery life estimation of Li-ion cell | |
| CN115994441A (en) | Big data cloud platform online battery life prediction method based on mechanism information | |
| CN114384435A (en) | WSA-LSTM algorithm-based self-adaptive prediction method for residual service life of new energy automobile power battery | |
| CN115236522A (en) | End-to-end capacity estimation method of energy storage battery based on hybrid deep neural network | |
| CN115201686A (en) | Lithium ion battery health state assessment method under incomplete charging and discharging data | |
| CN113449919B (en) | Power consumption prediction method and system based on feature and trend perception | |
| CN111337833B (en) | Lithium battery capacity integrated prediction method based on dynamic time-varying weight | |
| CN116449218B (en) | Lithium battery health state estimation method | |
| US20230305073A1 (en) | Method and apparatus for providing a predicted aging state of a device battery based on a predicted usage pattern | |
| Zhang et al. | Remaining useful life prediction of lithium-ion batteries based on TCN-DCN fusion model combined with IRRS filtering | |
| CN113884936B (en) | ISSA coupling DELM-based lithium ion battery health state prediction method | |
| CN115598546A (en) | Combined estimation method for SOH, SOC and RUL of lithium ion battery | |
| Uddin et al. | State of health estimation of lithium-ion batteries in vehicle-to-grid applications using recurrent neural networks for learning the impact of degradation stress factors |
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 |