CN117169743A - Battery health state estimation method and device based on partial data and model fusion - Google Patents
Battery health state estimation method and device based on partial data and model fusion Download PDFInfo
- Publication number
- CN117169743A CN117169743A CN202311124122.7A CN202311124122A CN117169743A CN 117169743 A CN117169743 A CN 117169743A CN 202311124122 A CN202311124122 A CN 202311124122A CN 117169743 A CN117169743 A CN 117169743A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- battery
- training
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 230000036541 health Effects 0.000 title claims abstract description 56
- 230000004927 fusion Effects 0.000 title claims abstract description 49
- 238000012549 training Methods 0.000 claims abstract description 55
- 238000007600 charging Methods 0.000 claims abstract description 53
- 238000012795 verification Methods 0.000 claims abstract description 15
- 239000012634 fragment Substances 0.000 claims abstract description 10
- 238000010276 construction Methods 0.000 claims abstract description 8
- 238000000605 extraction Methods 0.000 claims abstract description 5
- 230000032683 aging Effects 0.000 claims description 16
- 238000002790 cross-validation Methods 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 11
- 238000012216 screening Methods 0.000 claims description 9
- 238000010801 machine learning Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000010219 correlation analysis Methods 0.000 claims description 2
- 238000002474 experimental method Methods 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 claims description 2
- 238000012706 support-vector machine Methods 0.000 claims description 2
- 238000007500 overflow downdraw method Methods 0.000 abstract description 23
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 description 3
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 3
- 238000006731 degradation reaction Methods 0.000 description 3
- 229910052744 lithium Inorganic materials 0.000 description 3
- 229910001416 lithium ion Inorganic materials 0.000 description 3
- 230000015556 catabolic process Effects 0.000 description 2
- 210000004027 cell Anatomy 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000010277 constant-current charging Methods 0.000 description 1
- 238000010281 constant-current constant-voltage charging Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003487 electrochemical reaction Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- -1 nickel cobalt aluminum Chemical compound 0.000 description 1
- KJONHKAYOJNZEC-UHFFFAOYSA-N nitrazepam Chemical compound C12=CC([N+](=O)[O-])=CC=C2NC(=O)CN=C1C1=CC=CC=C1 KJONHKAYOJNZEC-UHFFFAOYSA-N 0.000 description 1
- 239000007774 positive electrode material Substances 0.000 description 1
- 230000008929 regeneration Effects 0.000 description 1
- 238000011069 regeneration method Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
Landscapes
- Secondary Cells (AREA)
Abstract
The estimation method comprises 4 steps of feature extraction and training set construction based on partial charging data, training a base model based on cross verification, generating a secondary data set, training a second layer model to fuse a plurality of base models, and on-line SOH fusion estimation; according to the method, health features are extracted from part of charging fragment data only, so that the practicability of the SOH estimation model is improved; meanwhile, by fusing a plurality of heterogeneous base models by using a second layer model, the automatic weight distribution of the base models is realized; generating a data set based on cross verification ensures generalization of the model; the SOH estimation precision and the robustness of the fusion method are obviously superior to those of a single base model, and the fusion method has important engineering application value; the invention also provides equipment, a system and a medium based on the method.
Description
Technical Field
The invention relates to the technical field of battery management of electric vehicles, energy storage stations or data centers, in particular to a battery health state estimation method and device based on partial data and model fusion.
Background
The lithium ion battery inevitably undergoes capacity degradation during use. The capacity degradation of a battery is typically strongly time-varying, non-linear due to complex internal electrochemical reactions and aging mechanisms. However, the capacity cannot be directly measured as the internal state of the battery. The state of health (SOH) of a battery is generally defined as the percentage of the current maximum capacity of the battery to the nominal capacity. Accurate and reliable battery SOH estimation is of great importance to the design of battery management strategies.
By virtue of the characteristics of flexibility, universality, easiness in implementation and the like, the battery SOH estimation method based on data driving becomes an effective solution. Data driven methods typically require efficient health features to be extracted during battery charging to build the model. For example, patent application publication No. CN202210504737.1 by Ji Dongxu et al discloses a method for estimating the state of health of a lithium battery based on neural network and transfer integrated learning, which uses complete constant-current constant-voltage charging data to estimate the state of health of the battery; however, in practical applications, the initial state of charge of the battery is random. To maximize the practicality of the current data driving method, it is necessary to implement SOH estimation with only partial charge data. In addition, existing data-driven methods typically use a single model for SOH estimation, which has limited prediction accuracy and robustness. More importantly, a single model does not achieve optimal results on all test data due to the inherent limitations of the model. While some methods have proposed multi-model integration to estimate SOH, for example, state of the art Shen et al ("Deep convolutional neural networks with ensemble learning and transfer learning for capacity estimation of lithium-ion batteries," appl. Energy, vol.260, 2020.) integrate multiple deep-nerves together a network through neuron weights to estimate battery SOH; battery SOH fusion estimation was achieved by Cheng et al ("An ensemble prognostic method for lithium-ion battery capacity estimation based on time-varying weight allocation," appl.energy, vol.266, 2020.) at the university of Beijing aerospace, based on inducing an ordered weighted average operator to assign time-varying weights to each base model. However, these existing methods still face the problems of single base model, complex weight calculation, poor generalization and the like. These problems all greatly affect the robustness of the SOH estimation model in practical use.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a battery health state estimation method and equipment based on partial data and model fusion, which comprises 4 steps of feature extraction and training set construction based on partial charging data, training a base model based on cross verification and generating a secondary data set, training a second layer model to fuse a plurality of base models and on-line SOH fusion estimation; according to the method, health features are extracted from part of charging fragment data only, so that the practicability of the SOH estimation model is improved; meanwhile, by fusing a plurality of heterogeneous base models by using a second layer model, the automatic weight distribution of the base models is realized; cross-validation based dataset generation ensures generalization of the model. The SOH estimation precision and the robustness of the fusion method are obviously superior to those of a single model, and the fusion method has important engineering application value.
In order to achieve the above purpose, the technical scheme of the invention is as follows:
the battery state of health estimation method based on partial data and model fusion comprises the following steps:
step (1), extracting health characteristics based on partial charging data and constructing a training set
Firstly, screening charging data based on battery cycle aging experimental data; then, selecting segment voltage data for modeling, and extracting statistical characteristics of charging time and charging voltage based on the data; finally, integrating the statistical characteristics of the data of each cycle segment and the corresponding SOH value of the battery to complete the construction of a training data set;
step (2) training the base model based on cross-validation and generating a secondary dataset
Firstly, selecting a plurality of heterogeneous machine learning models as a base model and determining data division scores for cross verification; then, based on the training set constructed in the step (1), training and saving a plurality of base models are completed; finally, splicing the predicted value of each base model subjected to the folding cross verification with the corresponding SOH true value to form a secondary training data set;
step (3), training a second layer model to fuse a plurality of base models
Training a second layer model based on the second training data set generated in the step (2), automatically distributing weights to the plurality of base models in the step (2), namely realizing fusion of predicted values of the plurality of base models, and storing the trained second layer model;
step (4), battery SOH on-line fusion estimation
Firstly, extracting segment charging voltage data determined in the step (1) in the charging process of a battery, and extracting health features corresponding to the segment data; then, the base model trained in the step (2) is called to predict, and a plurality of predicted values are obtained; and finally, inputting the predicted values of the plurality of base models into the second layer model in the step (3), outputting to obtain a final estimated value of the SOH of the battery, and finishing the online fusion estimation of the SOH of the battery.
An apparatus for a battery state of health estimation method based on partial data and model fusion includes a memory and a processor;
a memory: storing the computer program of the battery state of health estimation method based on partial data and model fusion, which is a computer readable device;
a processor: the battery state of health estimation method is used for executing the battery state of health estimation method based on the partial data and the model fusion.
A system of a battery state of health estimation method based on partial data and model fusion comprises a processing instruction module for executing each step in the battery state of health estimation method based on partial data and model fusion.
A computer readable storage medium storing a computer program which, when executed by a processor, enables the battery state of health estimation method based on partial data and model fusion.
Compared with the prior art, the technical scheme provided by the invention has the following technical effects:
(1) And health features are extracted only by relying on part of charging data, so that the practicability of the SOH estimation model is improved.
(2) The fusion method automatically integrates a plurality of heterogeneous base models through a second layer model, so that automatic weight distribution of the base models is realized, and generalization of the models is ensured based on cross-validation data set generation. The SOH estimation precision and the robustness of the fusion method are obviously superior to those of a single model.
(3) Taking NASA aging dataset as an example, the SOH estimation average absolute error of the battery was only 0.42% after using the fusion method. SOH estimation accuracy was improved by 65%, 32%, 70% and 76% compared to the four base models, respectively.
In summary, the health features are extracted from only part of the charging fragment data, so that the practicability of the SOH estimation model is improved; meanwhile, by fusing a plurality of heterogeneous base models by using a second layer model, the automatic weight distribution of the base models is realized; generating a data set based on cross verification ensures generalization of the model; the SOH estimation precision and the robustness of the fusion method are obviously superior to those of a single base model, and the fusion method has important engineering application value.
Drawings
Fig. 1 is a schematic flow chart of a method for estimating SOH of a lithium battery based on partial charge data and model stack fusion according to the present invention.
FIG. 2 is a graph of aging trace and charge of NCA cells in a verification case of the present invention; fig. 2 (a) is a schematic diagram of an aging trace, and fig. 2 (b) is a charging graph.
Fig. 3 is a three-dimensional view of health features extracted based on partial charge data and battery capacity in a verification case of the present invention.
FIG. 4 is a diagram of the final estimation result of the multi-model stacking fusion SOH estimation method in the verification case of the present invention; fig. 4 (a) is a multi-model capacity estimation result comparison chart, and fig. 4 (b) is a capacity estimation error comparison chart.
Detailed Description
The technical scheme of the invention is further described in detail below with reference to the accompanying drawings:
this invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
As shown in fig. 1, the invention provides a battery state of health estimation method based on partial data and model fusion, which comprises the following steps:
and (1) extracting health characteristics based on the partial charging data and constructing a training set. Firstly, screening charging data based on battery cycle aging experimental data; then, selecting segment voltage data for modeling, and extracting statistical characteristics of charging time and charging voltage based on the data; and finally, integrating the statistical characteristics of the data of each cycle segment and the corresponding SOH value of the battery to complete the construction of the training data set.
The specific implementation mode is as follows:
step 1.1, screening charging process data of a battery cycle aging experiment, analyzing the correlation between fragment data and a battery SOH value according to a correlation coefficient, for example, measuring the correlation by using methods such as a Pelson coefficient, a Szelman coefficient or gray correlation analysis, and determining fragment voltage data with higher correlation for modeling by combining with an actual application scene.
And 1.2, calculating statistical characteristics related to charging time, charging voltage and charging temperature based on the selected segment voltage data, wherein the statistical characteristics comprise a voltage mean value, a voltage variance, a charging time maximum value, a charging time average value and a charging temperature average value, and extracting the statistical characteristics as health characteristics.
And 1.3, integrating the health features extracted in the step 1.2 with the SOH value of the battery corresponding to each cycle to complete the construction of a training data set for subsequent SOH estimation modeling.
And (2) training a base model based on the cross-validation and generating a secondary data set. Firstly, selecting a plurality of heterogeneous machine learning models as a base model and determining data division scores for cross verification; then, based on the training set constructed in the step (1), training and saving a plurality of base models are completed; and finally, splicing the predicted value of each base model subjected to the fold cross verification with the corresponding SOH true value to form a secondary training data set. The specific embodiments are as follows:
step 2.1, selecting a base model and determining the cross-validation fold number; according to actual application scenes and hardware computing capacity, selecting a plurality of heterogeneous machine learning models as base models, wherein the machine learning models comprise Gaussian process regression, a support vector machine, an extreme learning machine, a neural network and the like; the data partitioning fold number for cross-validation is determined, typically the fold number is taken as 5 or 10.
Step 2.2, training a base model and storing; based on the training data set in the step 1.3, training the plurality of base models selected in the step 2.1 in a cross-validation mode, and storing the models after training is completed.
Step 2.3, generating a secondary data set; in the cross-validation modeling in step 2.2, for each base model, preserving the predicted value of each base model and longitudinally stitching them together to form 1 column of feature data; then, transversely splicing the prediction data of the plurality of base models together to form a plurality of columns of characteristic data; and finally, integrating the characteristic data with the real SOH value of the battery corresponding to each aging cycle to generate a secondary training data set for training a subsequent secondary model.
Training a second layer model to fuse a plurality of base models; training a second-layer model based on the second-level training data set generated in the step (2), fusing the predictions of the plurality of base models in the step (2), and storing the trained second-layer model.
The specific embodiments are as follows:
step 3.1, selecting a second layer model for preventing overfitting according to an actual application scene, wherein the second layer model selects a linear model;
step 3.2, training the second layer model selected in the step 3.1 based on the second training data set generated in the step 2.3, and automatically distributing weights to the base models in the step 2.1 by using the second layer model, namely realizing fusion of predicted values of a plurality of base models;
and 3.3, storing the second layer model trained in the step 3.2 locally so as to be convenient to call in time during online application.
And (4) carrying out on-line fusion estimation on the SOH of the battery. Firstly, extracting segment charging voltage data determined in the step (1) in the charging process of a battery, and extracting health features corresponding to the segment data; then, the base model trained in the step (2) is called to predict, and a plurality of predicted values are obtained; and finally, inputting the predicted values of the plurality of base models into the second layer model in the step (3), outputting to obtain a final estimated value of the SOH of the battery, and finishing the online fusion estimation of the SOH of the battery.
The specific embodiments are as follows:
step 4.1, screening charging data and calculating health characteristics; under a real test scene, monitoring battery charging process data, and screening out fragment voltage data determined in the step 1.1; calculating the characteristic quantity corresponding to the charging data of the section according to the characteristic extraction method in the step 1.2;
step 4.2, calling a base model to predict; calling the plurality of base models stored in the step 2.2, and predicting the characteristic quantity in the step 4.1 to obtain predicted values of the plurality of base models;
step 4.3, calling a two-layer model to predict to realize online SOH estimation; and (3) inputting the predicted values of the plurality of base models in the step (4.2) into the second layer model stored offline in the step (3.3) for prediction, and outputting to obtain a final SOH predicted value, namely completing the online fusion estimation of the SOH of the battery.
An apparatus for a battery state of health estimation method based on partial data and model fusion includes a memory and a processor;
a memory: storing the computer program of the battery state of health estimation method based on partial data and model fusion, which is a computer readable device;
a processor: the battery state of health estimation method is used for executing the battery state of health estimation method based on the partial data and the model fusion.
A system of a battery state of health estimation method based on partial data and model fusion comprises a processing instruction module for executing each step in the battery state of health estimation method based on partial data and model fusion.
A computer readable storage medium storing a computer program which, when executed by a processor, enables the battery state of health estimation method based on partial data and model fusion.
In summary, the invention mainly discloses a novel lithium battery health state estimation method based on partial charging data and model stacking fusion, which mainly comprises 4 steps of feature extraction and training set construction based on partial charging data, training a base model based on cross verification, generating a secondary data set, training a second layer model to fuse a plurality of base models, and on-line SOH fusion estimation. According to the method, the health features are extracted from part of the charging fragment data only, so that the practicability of the SOH estimation model is improved. Meanwhile, the weight of the base model is automatically distributed by fusing a plurality of heterogeneous base models by using a second layer model. Cross-validation based dataset generation ensures generalization of the model. The SOH estimation precision and the robustness of the fusion method are obviously superior to those of a single model, and the fusion method has important engineering application value.
In order to verify the feasibility and effectiveness of the method, on-line SOH estimation is performed by taking a #5 battery (the positive electrode material is a nickel cobalt aluminum NCA ternary material) as a test set example based on an NASA battery aging data set, and the estimation process and the estimation result are visualized.
As shown in fig. 2, a plot of capacity aging trace and charge for the NASA dataset #5 battery is shown. The battery capacity trace has obvious nonlinear degradation process accompanied with capacity regeneration phenomenon. The constant-current charging voltage curve of the battery gradually moves leftwards along with the aging deepening, and the charging time, namely the charging capacity, is reduced. In practical use of the battery, complete charge data is difficult to obtain. Therefore, it is necessary to perform SOH estimation using the charge data extraction features of the segments.
Fig. 3 shows 3 health characteristics closely related to battery aging, namely, a segment charging time (F1), a voltage average (F2) and a voltage standard deviation (F3), extracted from segment data of a charging voltage interval of #5 battery [3.9,4.1 ]. It can be seen from fig. 3 that the three features maintain a strong correlation with SOH, which is well suited for indicating battery aging, which is the basis for accurate estimation of battery SOH.
As shown in fig. 4, the SOH estimation result and the error map of the #5 battery are shown. The first 80% of the data for #5 cells was used for training and the last 20% for testing. Fig. 4 (a) shows a comparison of SOH estimation results of the fusion method and the four base models. It can be observed that the estimates of the Linear Regression (LR) model and the Gaussian Process Regression (GPR) model are almost always smaller, while the estimates of the Extreme Learning Machine (ELM) model are almost always larger. This phenomenon reveals unstable SOH estimation performance using a single model, and may cause a great SOH estimation error in practical applications. The SOH estimated by the Fusion method (Fusion) is closest to a real SOH decay curve, which shows that the Fusion method can well fuse the advantages of each basic model and reduce SOH estimation errors. The SOH curve estimated by the fusion method is closest to the real SOH decay curve. Fig. 4 (b) shows a fusion method and SOH estimation error comparison of four base models for a capacity estimation error comparison graph, and it can be seen that the SOH estimation error of the fusion method is minimum. It is noted that in the shaded area in the figure, the base model has a very large SOH estimation error (more than 4%), and the fusion method can still control the estimation error to about 2%, which indicates that the fusion method has very strong robustness. Using Mean Absolute Error (MAE) as an evaluation index, MAE of the four base models was 1.21%, 0.62%, 1.41%, 1.76%, respectively, and fusion method was 0.42%. Compared with four base models, the SOH estimation accuracy of the fusion method is improved by 65%, 32%, 70% and 76% respectively. The above results confirm the advantages of the fusion method of the present invention. The method can effectively solve the problem of unstable performance of a single data driving model, and by integrating a plurality of heterogeneous model estimation, the SOH estimation precision and robustness can be greatly improved by the fusion method, the confidence of data model popularization is improved, and the method has a wide application prospect.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
While the foregoing is directed to embodiments of the present invention, other and further details of the invention may be had by the present invention, it should be understood that the foregoing description is merely illustrative of the present invention and that no limitations are intended to the scope of the invention, except insofar as modifications, equivalents, improvements or modifications are within the spirit and principles of the invention.
Claims (8)
1. The battery state of health estimation method based on partial data and model fusion is characterized by comprising the following steps:
step (1), extracting health characteristics based on partial charging data and constructing a training set
Firstly, screening charging data based on battery cycle aging experimental data; then, selecting segment voltage data for modeling, and extracting statistical characteristics of charging time and charging voltage based on the data; finally, integrating the statistical characteristics of the data of each cycle segment and the corresponding SOH value of the battery to complete the construction of a training data set;
step (2) training the base model based on cross-validation and generating a secondary dataset
Firstly, selecting a plurality of heterogeneous machine learning models as a base model and determining data division scores for cross verification; then, based on the training set constructed in the step (1), training and saving a plurality of base models are completed; finally, splicing the predicted value of each base model subjected to the folding cross verification with the corresponding SOH true value to form a secondary training data set;
step (3), training a second layer model to fuse a plurality of base models
Training a second layer model based on the second training data set generated in the step (2), automatically distributing weights to the plurality of base models in the step (2), namely realizing fusion of predicted values of the plurality of base models, and storing the trained second layer model;
step (4), battery SOH on-line fusion estimation
Firstly, extracting segment charging voltage data determined in the step (1) in the charging process of a battery, and extracting health features corresponding to the segment data; then, the base model trained in the step (2) is called to predict, and a plurality of predicted values are obtained; and finally, inputting the predicted values of the plurality of base models into the second layer model in the step (3), outputting to obtain a final estimated value of the SOH of the battery, and finishing the online fusion estimation of the SOH of the battery.
2. The method for estimating a battery state of health based on partial data and model fusion according to claim 1, wherein step (1) is specifically:
step 1.1, screening charging process data of a battery cycle aging experiment, analyzing the correlation between fragment data and a battery SOH value according to a correlation coefficient (correlation is measured by using methods such as a Pierson coefficient, a Szelman coefficient or gray correlation analysis), and determining fragment voltage data with higher correlation for modeling by combining with an actual application scene;
step 1.2, calculating statistical characteristics related to charging time, charging voltage and charging temperature based on the selected segment voltage data, wherein the statistical characteristics comprise a voltage average value, a voltage variance, a charging time maximum value, a charging time average value and a charging temperature average value, and extracting the statistical characteristics as health characteristics;
and 1.3, integrating the health features extracted in the step 1.2 with the SOH value of the battery corresponding to each cycle to complete the construction of a training data set for subsequent SOH estimation modeling.
3. The method for estimating a battery state of health based on partial data and model fusion according to claim 2, wherein step (2) is specifically:
step 2.1, selecting a base model and determining the cross-validation fold number; according to actual application scenes and hardware computing capacity, selecting a plurality of heterogeneous machine learning models as base models, wherein the machine learning models comprise Gaussian process regression, a support vector machine, an extreme learning machine, a neural network and the like; determining the data division fold number for cross verification, wherein the general fold number is 5 or 10;
step 2.2, training a base model and storing; training the plurality of base models selected in the step 2.1 in a cross-validation mode based on the training data set in the step 1.3, and storing the models after training is completed;
step 2.3, generating a secondary data set; in the cross-validation modeling in step 2.2, for each base model, preserving the predicted value of each base model and longitudinally stitching them together to form 1 column of feature data; then, transversely splicing the prediction data of the plurality of base models together to form a plurality of columns of characteristic data; and finally, integrating the characteristic data with the real SOH value of the battery corresponding to each aging cycle to generate a secondary training data set for training a subsequent secondary model.
4. The method for estimating a battery state of health based on partial data and model fusion according to claim 3, wherein step (3) is specifically:
step 3.1, selecting a second layer model for preventing overfitting according to an actual application scene, wherein the second layer model selects a linear model;
step 3.2, training the second layer model selected in the step 3.1 based on the second training data set generated in the step 2.3, and automatically distributing weights to the base models in the step 2.1 by using the second layer model, namely realizing fusion of predicted values of a plurality of base models;
and 3.3, storing the second layer model trained in the step 3.2 locally so as to be convenient to call in time during online application.
5. The method for estimating a battery state of health based on partial data and model fusion according to claim 4, wherein step (4) is specifically:
step 4.1, screening charging data and calculating health characteristics; under a real test scene, monitoring battery charging process data, and screening out fragment voltage data determined in the step 1.1; calculating the characteristic quantity corresponding to the charging data of the section according to the characteristic extraction method in the step 1.2;
step 4.2, calling a base model to predict; calling the plurality of base models stored in the step 2.2, and predicting the characteristic quantity in the step 4.1 to obtain predicted values of the plurality of base models;
step 4.3, calling a two-layer model to predict to realize online SOH estimation; and (3) inputting the predicted values of the plurality of base models in the step (4.2) into the second layer model stored offline in the step (3.3) for prediction, and outputting to obtain a final SOH predicted value, namely completing the online fusion estimation of the SOH of the battery.
6. An apparatus for a battery state of health estimation method based on partial data and model fusion includes a memory and a processor; it is characterized in that the method comprises the steps of,
a memory: a computer program for storing the battery state of health estimation method based on partial data and model fusion as set forth in claim 5, as a computer-readable device;
a processor: the battery state of health estimation method is used for executing the battery state of health estimation method based on the partial data and the model fusion.
7. A system for a battery state of health estimation method based on partial data and model fusion, comprising a processing instruction module for performing each step of the battery state of health estimation method based on partial data and model fusion as claimed in claim 5.
8. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program, which when executed by a processor is capable of implementing the battery state of health estimation method based on partial data and model fusion of claim 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311124122.7A CN117169743A (en) | 2023-09-01 | 2023-09-01 | Battery health state estimation method and device based on partial data and model fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311124122.7A CN117169743A (en) | 2023-09-01 | 2023-09-01 | Battery health state estimation method and device based on partial data and model fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117169743A true CN117169743A (en) | 2023-12-05 |
Family
ID=88933089
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311124122.7A Pending CN117169743A (en) | 2023-09-01 | 2023-09-01 | Battery health state estimation method and device based on partial data and model fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117169743A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118068199A (en) * | 2024-04-18 | 2024-05-24 | 无锡锂云科技有限公司 | Battery charge-discharge curve prediction method and device, electronic equipment and storage medium |
-
2023
- 2023-09-01 CN CN202311124122.7A patent/CN117169743A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118068199A (en) * | 2024-04-18 | 2024-05-24 | 无锡锂云科技有限公司 | Battery charge-discharge curve prediction method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xiong et al. | An electrochemical model based degradation state identification method of Lithium-ion battery for all-climate electric vehicles application | |
| Kim et al. | Estimation of Li-ion battery state of health based on multilayer perceptron: As an EV application | |
| Mamo et al. | Long short-term memory with attention mechanism for state of charge estimation of lithium-ion batteries | |
| Wang et al. | Health diagnosis for lithium-ion battery by combining partial incremental capacity and deep belief network during insufficient discharge profile | |
| CN114035098A (en) | Lithium battery health state prediction method integrating future working condition information and historical state information | |
| Jafari et al. | Prediction of the battery state using the digital twin framework based on the battery management system | |
| CN117169743A (en) | Battery health state estimation method and device based on partial data and model fusion | |
| Ouyang et al. | Prognostics and health management of lithium-ion batteries based on modeling techniques and Bayesian approaches: A review | |
| CN116008844A (en) | Vehicle lithium battery health state online prediction method adapting to quick charge strategy | |
| Jiang et al. | Multiple health indicators assisting data-driven prediction of the later service life for lithium-ion batteries | |
| Wen et al. | Lithium battery health state assessment based on vehicle-to-grid (V2G) real-world data and natural gradient boosting model | |
| Cao et al. | A flexible battery capacity estimation method based on partial voltage curves and polynomial fitting | |
| CN113376541B (en) | Lithium ion battery health state prediction method based on CRJ network | |
| Chen et al. | State of health estimation of lithium-ion batteries based on equivalent circuit model and data-driven method | |
| Huang et al. | State of health estimation of lithium-ion batteries based on fine-tuning or rebuilding transfer learning strategies combined with new features mining | |
| CN112327169B (en) | Lithium battery residual life prediction method | |
| US20230349977A1 (en) | Method and apparatus for estimating state of health of battery | |
| Singh et al. | State of charge estimation techniques of Li-ion battery of electric vehicles | |
| Li et al. | A state-of-health estimation method of lithium-ion batteries using ICA and SVM | |
| CN116774045A (en) | Lithium battery health state prediction method based on HHO-SVR | |
| Zhang et al. | The early prediction of lithium-ion battery remaining useful life using a novel long short-term memory network | |
| CN111190112B (en) | Battery charging and discharging prediction method and system based on big data analysis | |
| CN115236521A (en) | Lithium ion battery health state estimation method based on semi-supervised cooperative training framework | |
| Xiong et al. | Improved cooperative competitive particle swarm optimization and nonlinear coefficient temperature decreasing simulated annealing-back propagation methods for state of health estimation of energy storage batteries | |
| Zheng et al. | Research on Data-Driven-Based Remaining Useful Life of Lithium-ion Battery |
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 |