CN113094994A - Power battery prediction method based on big data migration learning - Google Patents
Power battery prediction method based on big data migration learning Download PDFInfo
- Publication number
- CN113094994A CN113094994A CN202110388363.7A CN202110388363A CN113094994A CN 113094994 A CN113094994 A CN 113094994A CN 202110388363 A CN202110388363 A CN 202110388363A CN 113094994 A CN113094994 A CN 113094994A
- Authority
- CN
- China
- Prior art keywords
- power battery
- model
- data
- transfer learning
- training
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000013508 migration Methods 0.000 title claims abstract description 21
- 230000005012 migration Effects 0.000 title claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 72
- 238000013526 transfer learning Methods 0.000 claims abstract description 38
- 238000013527 convolutional neural network Methods 0.000 claims description 12
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000013145 classification model Methods 0.000 claims description 5
- 230000000306 recurrent effect Effects 0.000 claims description 4
- 125000004122 cyclic group Chemical group 0.000 claims description 2
- 238000011161 development Methods 0.000 abstract description 33
- 238000012546 transfer Methods 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 7
- 238000012544 monitoring process Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000178 monomer Substances 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Secondary Cells (AREA)
Abstract
The invention relates to a power battery prediction method based on big data migration learning, which comprises the following steps: establishing and training a transfer learning pre-training model with a plurality of reserved characteristics for predicting the power battery based on big data of the power battery; when the new type of power battery needs to be predicted, a part of time sequence data of the new type of power battery is utilized to finely adjust, train, transfer, learn and pre-train the model, part of features in the part of time sequence data of the new type of power battery are correspondingly applied to reserved features to obtain a new prediction model suitable for the new type of power battery, and then the new prediction model is utilized to predict the power battery to be predicted and obtain a prediction result. The method can accelerate the model development speed, solve the problem of model development when the data of a new power battery product is slightly less, solve the problem of non-uniform data characteristics of different power battery products, reduce the difficulty of model development and improve the efficiency of model development.
Description
Technical Field
The invention belongs to the technical field of power battery attribute calculation and management, and particularly relates to a power battery prediction method based on big data migration learning.
Background
A power battery of the new energy automobile can generate a large amount of monitoring data in the running process, and a machine learning prediction model can be established by using the data to predict various attributes of the power battery. For example, the highest temperature of the power battery in a future period is predicted by using historical monitoring data of the power battery, and the control strategy of the vehicle is changed according to the temperature, so that the temperature can be controlled within a safe range, and the risk of thermal runaway of the power battery is reduced.
The process of establishing a machine learning, especially a deep learning prediction model by using monitoring data comprises the steps of preparing a large number of data sets, establishing a model structure, training the model by using the data sets, further adjusting parameters according to training results, and continuously repeating the training and parameter adjusting processes until a satisfactory prediction effect is finally obtained.
As shown in fig. 1, the conventional model development process is:
1) power battery timing data
The time sequence data used here is a time sequence segment intercepted from a large segment of complete power battery data by a sliding window method, usually 1 hour of segment data is intercepted, for example, if the data sampling frequency is higher, shorter segment data can be used, but in order to ensure the prediction effect, the data length should not be less than 30 minutes.
2) Data cleansing
The purpose of data cleansing is to cleanse missing or abnormal values in the data:
loss value: filling the data by using a mean value, a median value or a neighbor value for slight data loss (for example, few voltage values in 95 voltage values are lost), and directly deleting the data for serious data loss (for example, most of 95 voltage values are lost);
abnormal value: and judging whether the data is in a normal range or not by using statistical judgment or domain knowledge, and replacing the data in an abnormal range in a specific processing mode similar to missing value filling by using a mean value, a median value or a neighboring value.
3) Data normalization
In order to eliminate dimension difference among different characteristics, a min-max normalization method is used for carrying out normalization processing on data.
4) Data size normalization
In order to facilitate the data processing by the CNN method, the data size is standardized, but because the adjacent features of the power battery data do not have the neighborhood relation in time and space, and the data width direction has a uniform size, the data does not need to be scaled in the width direction. In the height direction, because the data has a neighborhood relationship in time, scaling can be performed in the height direction of the data so that different input data maintain a uniform standard size.
5) Training model
And repeatedly training the model for a long time by using mass data, detecting the prediction effect of the model on the verification set in the training process, and storing the model after the model is converged to obtain the trained model.
The new energy automobile is a rapidly developing industry, new products are launched rapidly, the rapid development brings many opportunities for creating a machine learning model, and simultaneously brings many challenges, which are represented by many product types, non-uniform monitoring data formats (for example, the number of monomer voltages in different automobile models may be different), less data accumulation of new products, and the like, which all increase the difficulty of model development.
Disclosure of Invention
The invention aims to provide a power battery prediction method based on big data migration learning, which can reduce the difficulty of model development and improve the efficiency of model development.
In order to achieve the purpose, the invention adopts the technical scheme that:
a power battery prediction method based on big data migration learning comprises the following steps:
establishing and training a transfer learning pre-training model with a plurality of reserved characteristics for predicting the power battery in advance based on big data of the power battery;
when a new type of power battery needs to be predicted, utilizing partial time sequence data of the new type of power battery to finely train the transfer learning pre-training model, and correspondingly applying partial features in the partial time sequence data of the new type of power battery to the reserved features to obtain a new prediction model suitable for the new type of power battery; and when the power battery to be predicted, which belongs to the new type of power battery, is predicted, predicting the power battery to be predicted by using the new prediction model and obtaining a prediction result.
The default value of the reserved feature in the transfer learning pre-training model is 0.
The transfer learning pre-training model uses a convolutional neural network structure or a cyclic neural network structure.
When the migration learning pre-training model uses a convolutional neural network structure, the migration learning pre-training model adopts a VGG series, a GoogLeNet series, a ResNet series or a DenseNet series model.
When the transfer learning pre-training model uses a recurrent neural network structure, the transfer learning pre-training model adopts an LSTM or GRU model.
The transfer learning pre-training model is a regression model or a classification model.
Due to the application of the technical scheme, compared with the prior art, the invention has the following advantages: the method can accelerate the model development speed, solve the problem of model development when the data of a new power battery product is slightly less, solve the problem of non-uniform data characteristics of different power battery products, reduce the difficulty of model development and improve the efficiency of model development.
Drawings
FIG. 1 is a flow chart of a conventional development process.
FIG. 2 is a flow chart of development of a transfer learning pre-training model in the big data transfer learning-based power battery prediction method of the invention.
FIG. 3 is a flow chart of development of a new prediction model in the big data transfer learning-based power battery prediction method.
Detailed Description
The invention will be further described with reference to examples of embodiments shown in the drawings to which the invention is attached.
The first embodiment is as follows: a power battery prediction method based on big data migration learning comprises the following steps: establishing and training a transfer learning pre-training model with a plurality of reserved characteristics for predicting the power battery in advance based on big data of the power battery; when the new type of power battery needs to be predicted, a part of time sequence data of the new type of power battery is utilized to finely adjust, train, migrate, learn and pre-train the model, and part of features in the part of time sequence data of the new type of power battery are correspondingly applied to reserved features, so that a new prediction model suitable for the new type of power battery is obtained; and when the power battery to be predicted, which belongs to the new type of power battery, is predicted, predicting the power battery to be predicted by using the new prediction model and obtaining a prediction result. In short, the big data transfer learning-based power battery prediction method comprises two parts, namely creating a transfer learning pre-training model and creating a new prediction model by using the transfer learning pre-training model.
1. Development process of transfer learning pre-training model
1) Development process
The development process of the transfer learning pre-training model is basically the same as the development process of the conventional model, and as shown in fig. 2, the difference is that a plurality of reserved features are added when determining the input features of the model. Because the new energy automobile enterprises are numerous and each enterprise has a plurality of automobile models, and the characteristics of the power battery data are inevitably different due to different configuration of the automobile models in spite of the specification of the national standard format. To cope with this problem, the present solution uses a sufficient number of features to ensure consistency in data characteristics by increasing the method of reserving features. For example, the number of features of the single power battery is usually about 100, the number of the reserved features can be increased to make the total number of the features of the single battery reach 120, and the default value of the reserved features in the migration learning pre-training model is 0.
The reservation features mainly include 3 types: a battery cell reservation characteristic, a battery pack temperature reservation characteristic, and other characteristic reservation characteristics.
2) Model network structure
The transfer learning pre-training model uses a Convolutional Neural Network (CNN) structure or a Recurrent Neural Network (RNN) structure. When the migration learning pre-training model uses a convolutional neural network structure, the migration learning pre-training model adopts a model of VGG series, GoogLeNet series, ResNet series or DenseNet series. When the transfer learning pre-training model uses a recurrent neural network structure, the transfer learning pre-training model adopts an LSTM or GRU model.
In contrast, the CNN method model has higher flexibility during the fine tuning training, so that the transfer learning pre-training model in this embodiment uses standard convolutional neural network structures, such as VGG series, google lenet series, ResNet series, DenseNet series, and the like, to obtain transfer learning pre-training models with various network structures, so as to meet the requirements of different service scenarios on prediction accuracy, speed, and the like.
Because the standard convolutional neural network is created for the visual field, in order to be used on power battery data, the structure of the standard convolutional neural network needs to be adjusted, which mainly includes:
a) changing the number of input channels in _ channels of the first layer of convolution kernel from 3 to 1;
b) modifying the output quantity of the FC layer: if the model is a regression model, changing the output dimension into a target dimension value and removing the softmax layer; in the case of the classification model, the FC output dimension is changed to the number of classes classified while preserving the softmax layer.
3) Type of pre-trained model
The migration learning pre-training model selects attributes capable of reflecting the overall condition to be used as a regression model or a classification model (for example, the highest temperature is used as the regression model or the overvoltage alarm is used as the classification model), a large data set is used for full training, parameter adjustment and retraining, so that the model can learn an effective data mode from the large data set, and finally the trained model is stored to be used as the migration learning pre-training model.
2. New prediction model development process based on transfer learning pre-training model
When a new type of power battery product, namely a new type of power battery demand to be predicted, is available, a proper pre-training model is selected according to the requirements on the operation speed and the prediction precision, the quantity of the model output is modified according to the prediction data dimension, and then a new prediction model suitable for the new type of power battery can be quickly obtained by training in a fine tuning mode.
The new model development process based on the migration learning is generally similar to the conventional model development process as a whole, as shown in fig. 3, but has fundamental differences, including:
1) small batch of time series data
The migration learning can obtain the effect which can be achieved by mass data in the conventional model development method by using small-batch data fine tuning training by means of a general effective mode in a pre-training model, and the requirement of model development on data volume is reduced.
2) Feature alignment
When the pre-training model is used, the feature sequence of the new model data is arranged according to the features (including the reserved features) of the pre-training model:
voltage alignment: filling the voltage characteristics of the new model data into the voltage characteristic positions of the pre-trained model, wherein the default value of 0 is used for the positions which are not used;
temperature alignment: aligning with the voltage, and filling the temperature characteristics of the new modulus data into the temperature characteristic position of the pre-training model, wherein the position which is not used uses a default value of 0;
alignment of other features: and filling other features according to corresponding feature positions in the pre-training model, filling a default value of 0 if the pre-training features which are not used exist, filling the positions of the reserved other features if the pre-training features which are not used exist in the pre-training model, and filling other reserved features which are not used by using the default value of 0.
3) Fine-tuning pre-training model
The transfer learning is completely different from the conventional model in training, long-time training of a large amount of data is not needed, and a new prediction model achieving the expected prediction effect can be quickly obtained only by using small-batch data of a new type of power battery to finely adjust the pre-training model.
According to the scheme, a migration learning idea in deep learning is used, the problem of non-uniformity of data formats is solved by using a reserved attribute mode, a model is trained on a large data set to learn a general expression mode in data, and then the general expression mode is finely tuned and migrated to other products by using a small amount of data, so that the problem of model development caused by multiple product types and insufficient data quantity is solved. The scheme is mainly improved in the following 3 aspects:
1) the problem of module development when the data volume of a new product is smaller is solved
When a new product model is developed, the problem of small data volume caused by insufficient data accumulation often exists, so that the model cannot learn an effective expression mode from a small amount of data, and the model effect cannot reach the expected prediction performance. The method uses a transfer learning mode to directly transfer the effective data mode which is learned from a large amount of data in the pre-training model, so that the requirement on data volume when a new product model is developed is reduced.
2) Solves the problem of non-uniform characteristics of different product data
The reserved characteristics are used, the problem of characteristic difference of power battery data of different products is solved, different power battery data can be unified to the same model, the universality of the model is improved, and the scale of model development is reduced.
3) Accelerating model development speed
The large data set pre-training model is used for learning a general effective mode in the power battery data, when a new product model is developed, the pre-training model can be used for obtaining a model with expected prediction performance only by a fine-tuning training mode without starting all the models from zero, so that the development speed is accelerated, and the prediction performance is guaranteed.
The scheme has the advantages that:
in the conventional deep learning model development, a large amount of data, long-time training and a large amount of parameters are required to obtain a model with a prediction capability meeting the requirement, and in actual operation, various limitations on technical complexity, time investment and data set often exist.
In the field of electric automobiles, new products are launched quickly, and the problem of data set is represented by insufficient data accumulation, so that a model with good enough prediction performance cannot be obtained only by using less data to train the model. The method adopts a transfer learning method, exerts the advantages of big data, fully trains the models on a big data set, enables the models to learn the effective modes in the power battery data, and then uses the pre-training models for developing various new models, so that the new models can transfer and use the effective modes in the pre-training models, the development difficulty is reduced, and the expected model effect can be obtained only by less data and short-time fine-tuning training.
The diversity of different vehicle product configurations also often means differences in data characteristics, so that data of different products cannot be used universally, so that models have to be developed separately for each product, and heavy burdens are imposed on model development and model deployment. This patent is through the mode of reserving the characteristic, provides unified data format for different motorcycle types, and these reserve the characteristic when not using, and the corresponding weight is nearly in 0 to do not play a role. Once the reserved features are needed, the weights corresponding to the reserved features can be updated rapidly only by fine tuning and training a new model, so that the reserved features play a role, and finally the problem that the features of different vehicle types are not uniform is solved.
The above embodiments are merely illustrative of the technical ideas and features of the present invention, and the purpose thereof is to enable those skilled in the art to understand the contents of the present invention and implement the present invention, and not to limit the protection scope of the present invention. All equivalent changes and modifications made according to the spirit of the present invention should be covered within the protection scope of the present invention.
Claims (6)
1. A power battery prediction method based on big data migration learning is characterized in that: the power battery prediction method based on big data migration learning comprises the following steps:
establishing and training a transfer learning pre-training model with a plurality of reserved characteristics for predicting the power battery in advance based on big data of the power battery;
when a new type of power battery needs to be predicted, utilizing partial time sequence data of the new type of power battery to finely train the transfer learning pre-training model, and correspondingly applying partial features in the partial time sequence data of the power battery to be newly typed to the reserved features to obtain a new prediction model suitable for the new type of power battery; and when the power battery to be predicted, which belongs to the new type of power battery, is predicted, predicting the power battery to be predicted by using the new prediction model and obtaining a prediction result.
2. The method for predicting the power battery based on big data transfer learning according to claim 1, wherein the method comprises the following steps: the default value of the reserved feature in the transfer learning pre-training model is 0.
3. The method for predicting the power battery based on big data transfer learning according to claim 1, wherein the method comprises the following steps: the transfer learning pre-training model uses a convolutional neural network structure or a cyclic neural network structure.
4. The method for predicting the power battery based on big data transfer learning according to claim 3, wherein the method comprises the following steps: when the migration learning pre-training model uses a convolutional neural network structure, the migration learning pre-training model adopts a VGG series, a GoogLeNet series, a ResNet series or a DenseNet series model.
5. The method for predicting the power battery based on big data transfer learning according to claim 3, wherein the method comprises the following steps: when the transfer learning pre-training model uses a recurrent neural network structure, the transfer learning pre-training model adopts an LSTM or GRU model.
6. The method for predicting the power battery based on big data transfer learning according to claim 1, wherein the method comprises the following steps: the transfer learning pre-training model is a regression model or a classification model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110388363.7A CN113094994A (en) | 2021-04-12 | 2021-04-12 | Power battery prediction method based on big data migration learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110388363.7A CN113094994A (en) | 2021-04-12 | 2021-04-12 | Power battery prediction method based on big data migration learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113094994A true CN113094994A (en) | 2021-07-09 |
Family
ID=76676506
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110388363.7A Pending CN113094994A (en) | 2021-04-12 | 2021-04-12 | Power battery prediction method based on big data migration learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113094994A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114506247A (en) * | 2022-03-16 | 2022-05-17 | 中国科学技术大学 | Power battery active heat management system based on cloud cooperative control |
| CN115114863A (en) * | 2022-08-23 | 2022-09-27 | 苏州清研精准汽车科技有限公司 | Battery pack Y capacitance prediction method and device, computer equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2018072083A (en) * | 2016-10-26 | 2018-05-10 | 株式会社東芝 | Energy consumption prediction device and energy consumption prediction method |
| CN109754068A (en) * | 2018-12-04 | 2019-05-14 | 中科恒运股份有限公司 | Transfer learning method and terminal device based on deep learning pre-training model |
| CN112036084A (en) * | 2020-08-28 | 2020-12-04 | 北京航空航天大学 | Similar product life migration screening method and system |
-
2021
- 2021-04-12 CN CN202110388363.7A patent/CN113094994A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2018072083A (en) * | 2016-10-26 | 2018-05-10 | 株式会社東芝 | Energy consumption prediction device and energy consumption prediction method |
| CN109754068A (en) * | 2018-12-04 | 2019-05-14 | 中科恒运股份有限公司 | Transfer learning method and terminal device based on deep learning pre-training model |
| CN112036084A (en) * | 2020-08-28 | 2020-12-04 | 北京航空航天大学 | Similar product life migration screening method and system |
Non-Patent Citations (3)
| Title |
|---|
| S. SHEN ETC: "Deep convolutional neural networks with ensemble learning and transfer learning for capacity estimation of lithium-ion batteries", APPLIED ENERGY, vol. 260, pages 3 * |
| 叶林海: "小样本条件下的锂电池健康状态估计研究", 中国优秀硕士学位论文全文数据库 工程科技辑, no. 07, pages 042 - 820 * |
| 王一全 等: "基于LSTM-DaNN的动力电池SOC估算方法", 储能科学与技术, vol. 9, no. 06, pages 1969 - 1975 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114506247A (en) * | 2022-03-16 | 2022-05-17 | 中国科学技术大学 | Power battery active heat management system based on cloud cooperative control |
| CN114506247B (en) * | 2022-03-16 | 2024-02-09 | 中国科学技术大学 | Active thermal management system of power battery controlled cooperatively by cloud |
| CN115114863A (en) * | 2022-08-23 | 2022-09-27 | 苏州清研精准汽车科技有限公司 | Battery pack Y capacitance prediction method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113094994A (en) | Power battery prediction method based on big data migration learning | |
| KR102439041B1 (en) | Method and apparatus for diagnosing defect of battery cell based on neural network | |
| CN111047085B (en) | Hybrid vehicle working condition prediction method based on meta-learning | |
| CN111190111B (en) | Electrochemical energy storage battery state of charge estimation method, device and system | |
| US20220373601A1 (en) | Battery Performance Prediction | |
| CN106427589A (en) | Electric car driving range estimation method based on prediction of working condition and fuzzy energy consumption | |
| CN112798961B (en) | Method for predicting remaining service life of power battery of electric automobile | |
| CN112277927B (en) | Hybrid electric vehicle energy management method based on reinforcement learning | |
| CN111420898A (en) | Retired battery sorting method and system applied by same | |
| WO2023226355A1 (en) | Dual-ion battery fault detection method and system based on multi-source perception | |
| WO2021052540A1 (en) | Condition value for rechargeable batteries | |
| WO2022144542A1 (en) | Method for predicting condition parameter degradation of a cell | |
| CN115146723A (en) | Electrochemical model parameter identification method based on deep learning and heuristic algorithm | |
| CN108733633A (en) | A kind of the unbalanced data homing method and device of sample distribution adjustment | |
| CN116930769A (en) | Lithium battery modeling method based on bidirectional generation type antagonistic neural network | |
| CN117131687A (en) | Electric automobile battery management method based on data analysis | |
| Zhang et al. | Remaining useful life prediction of lithium-ion batteries based on TCN-DCN fusion model combined with IRRS filtering | |
| CN113788007B (en) | Layered real-time energy management method and system | |
| CN110927584B (en) | Neural network-based battery life extension prediction method | |
| CN114545928A (en) | Heterogeneous vehicle queue control method based on self-triggering distributed predictive control | |
| CN114547969A (en) | EMD-MRVR-based multi-stress battery life prediction method | |
| Guan et al. | Research on the method of remaining useful life prediction of lithium-ion battery based on LSTM | |
| CN116572797A (en) | Battery management system control method, device, server and storage medium | |
| Yu et al. | A fusion method for capacity estimation of fast-charging lithium-ion battery under varied degradation mode | |
| Li et al. | Evaluation method for consistency of lithium-ion battery packs in electric vehicles based on the Mahalanobis-Taguchi system |
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 |