CN116609677B - Battery state estimation method - Google Patents
Battery state estimation method Download PDFInfo
- Publication number
- CN116609677B CN116609677B CN202310879687.XA CN202310879687A CN116609677B CN 116609677 B CN116609677 B CN 116609677B CN 202310879687 A CN202310879687 A CN 202310879687A CN 116609677 B CN116609677 B CN 116609677B
- Authority
- CN
- China
- Prior art keywords
- battery
- model
- neural network
- training
- deep neural
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000013528 artificial neural network Methods 0.000 claims abstract description 75
- 238000012549 training Methods 0.000 claims abstract description 66
- 238000004458 analytical method Methods 0.000 claims abstract description 38
- 230000032683 aging Effects 0.000 claims abstract description 16
- 238000012795 verification Methods 0.000 claims abstract description 4
- 238000005516 engineering process Methods 0.000 claims description 7
- 230000009467 reduction Effects 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 238000013138 pruning Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000012937 correction Methods 0.000 claims description 2
- 230000010354 integration Effects 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 230000036541 health Effects 0.000 abstract description 5
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 7
- 229910001416 lithium ion Inorganic materials 0.000 description 7
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 230000001965 increasing effect Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 235000008733 Citrus aurantifolia Nutrition 0.000 description 1
- 235000011941 Tilia x europaea Nutrition 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000004571 lime Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
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]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- 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
- G06N3/098—Distributed learning, e.g. federated 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
- G06N3/08—Learning methods
- G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Secondary Cells (AREA)
Abstract
The invention provides a battery state estimation method, which is characterized in that a deep neural network is designed based on a hybrid physical model of a battery, a comprehensive analysis layer is arranged between a hidden layer and an output layer of the deep neural network, the output of the hidden layer is comprehensively analyzed to fuse various physical characteristics and working principles of the battery, and training output results of all models are mutually verified and revised, wherein the trained battery state estimation model comprises an equivalent circuit model, a thermodynamic model and an aging model of the battery, and the battery estimation states which can be obtained simultaneously comprise the residual capacity, the health state, a charge-discharge curve and the temperature distribution of the battery. The battery state estimation method provided by the invention can carry out mutual verification revision on the data of each model and the estimated state through the comprehensive analysis layer, effectively improves the estimation accuracy of the trained battery state estimation model on the battery state, provides convenience for battery management of the electric automobile, and improves the operation reliability of the electric automobile.
Description
Technical Field
The invention relates to the technical field of automobile storage battery management, in particular to a battery state estimation method.
Background
The lithium ion battery is used as a clean power supply, is more and more popular because of high energy density, low self-discharge rate, small volume, long service life and low pollution degree, and is particularly widely applied to a plurality of fields such as electric automobiles, communication base stations and the like.
However, the capacity of the lithium ion battery can be continuously declined when the lithium ion battery circulates each time, the battery capacity reduction can influence the endurance of the electric automobile, if the battery capacity cannot be accurately monitored, the electric automobile can be anchored in advance, and the driving experience and the driving safety are influenced, so that the lithium ion battery capacity estimation and prediction have important significance for the running reliability guarantee of the electric automobile.
In the prior art, a battery state estimation model is trained based on a neural network, and the state of a battery is estimated and predicted according to the trained battery state estimation model, so that the method is a common method for estimating the battery state of an electric automobile, wherein a lithium ion battery model can be roughly divided into an empirical model, an equivalent circuit model and an electrochemical model, in practical application, one model is generally selected for prediction, and in order to improve the accuracy of prediction, the number of data samples trained by the model is generally increased to improve the accuracy of model prediction, however, the prediction accuracy of single model prediction is still insufficient.
Disclosure of Invention
Based on the above, the invention aims to provide a battery state estimation method so as to improve the accuracy of battery state estimation, provide convenience for battery management of an electric automobile and improve the operation reliability of the electric automobile.
One aspect of the present invention provides a battery state estimation method, including:
establishing a hybrid physical model of the battery according to physical characteristics and working principles of the battery to construct a deep neural network according to the hybrid physical model, wherein the hybrid physical model comprises an equivalent circuit model, a thermodynamic model and an aging model;
acquiring a data sample, and training the deep neural network according to the data sample to obtain a battery state estimation model;
acquiring current state data of a battery, wherein the current state data of the battery comprises voltage, current and temperature, and inputting the current state data of the battery into a battery state estimation model to output a battery estimation state through the battery state estimation model, and the estimation state comprises residual capacity, a health state, a charge-discharge curve and temperature distribution;
the deep neural network comprises a hidden layer, an output layer and a comprehensive analysis layer arranged between the hidden layer and the output layer, wherein the comprehensive analysis layer is used for comprehensively analyzing output items of all subnetworks in the hidden layer so as to integrate physical characteristics and working principles of the battery and enhance the accuracy of the battery state estimation model;
the output items of the hidden layer comprise data and model characteristics;
the data comprise physical quantity and state quantity of the battery, and the comprehensive analysis of the data by the comprehensive analysis layer comprises normalization, standardization, dimension reduction and clustering treatment;
the model comprises the equivalent circuit model, the thermodynamic model and the aging model after training, and the comprehensive analysis of the model by the comprehensive analysis layer comprises weighting, fusion, pruning and regularization treatment.
Optionally, the output item of the hidden layer further includes features, the features include space-time features, frequency domain features and statistical features characterizing internal features of the battery, and the comprehensive analysis of the features by the comprehensive analysis layer includes combining, transforming, selecting and noise reduction.
Optionally, before the step of training the deep neural network from the data samples, further comprising:
and migrating the weight data in the trained similar battery state estimation model to the deep neural network as initial weight data of the deep neural network.
Optionally, the step of training the deep neural network from the data samples further comprises:
and performing enhanced expansion on the data samples to increase the number of the data samples.
Optionally, the step of training the deep neural network from the data samples further comprises:
and simultaneously executing a plurality of learning training tasks in the deep neural network, wherein the plurality of learning training tasks at least comprise SOC estimation training, SOH estimation training and charge-discharge curve prediction training.
Optionally, the step of training the deep neural network from the data samples further comprises:
training the deep neural network according to data samples of various batteries and a meta-learning technology to obtain a deep neural network with learning capability;
training the deep neural network with learning capability according to the data sample of the target battery to obtain a battery state estimation model of the target battery.
Optionally, the step of training the deep neural network from the data samples further comprises: and improving the training speed of the deep neural network according to the hardware accelerator.
Optionally, the step of training the deep neural network from the data samples further comprises:
uploading training data of the deep neural network to a cloud according to cloud computing;
and training the deep neural network in real time according to edge calculation.
Optionally, the method further comprises: and extracting an interpretability feature according to the battery state estimation model, and obtaining an aging influence factor of the battery according to the interpretability feature.
Optionally, the interpretable feature is further provided to the comprehensive analysis layer, and comprehensive analysis of the interpretable feature by the comprehensive analysis layer includes integration, promotion, verification and correction processing.
The battery state estimation method provided by the invention is based on a hybrid physical model design depth neural network of a battery, and a comprehensive analysis layer is arranged between a hidden layer and an output layer of the depth neural network, the output of the hidden layer is comprehensively analyzed, so that various physical characteristics and working principles of the battery are fused, the training output results of all models are mutually verified and revised, wherein the trained battery state estimation model comprises an equivalent circuit model, a thermodynamic model and an aging model of the battery, the battery estimation states which can be obtained simultaneously comprise the residual capacity, the health state, a charge-discharge curve and the temperature distribution of the battery, the comprehensive analysis layer can be used for comprehensively analyzing the output items of the hidden layer, the mutual verification revision of the data of all models and the estimation states is realized, the efficiency fusion of the hybrid physical model is realized, the estimation accuracy of the trained battery state estimation model on the battery state is effectively improved, convenience is provided for battery management of an electric automobile, and the running reliability of the electric automobile is improved.
Drawings
Fig. 1 is a schematic flow chart of a battery state estimation method according to an embodiment of the invention.
The invention will be further described in the following detailed description in conjunction with the above-described figures.
Detailed Description
In order that the invention may be readily understood, a more complete description of the invention will be rendered by reference to the appended drawings. Several embodiments of the invention are presented in the figures. This invention may, however, 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.
It will be understood that when an element is referred to as being "mounted" on another element, it can be directly on the other element or intervening elements may also be present. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or intervening elements may also be present. The terms "vertical," "horizontal," "left," "right," and the like are used herein for illustrative purposes only.
Unless defined otherwise, all 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. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.
Referring to fig. 1, a main flow chart of a battery state estimation method according to an embodiment of the present invention is shown, which includes:
step S01: and establishing a hybrid physical model of the battery according to the physical characteristics and the working principle of the battery to construct a deep neural network according to the hybrid physical model, wherein the hybrid physical model comprises an equivalent circuit model, a thermodynamic model and an aging model.
Step S02: and acquiring a data sample, and training the deep neural network according to the data sample to obtain a battery state estimation model.
Step S03: and acquiring current state data of the battery, wherein the current state data of the battery comprises voltage, current and temperature, and inputting the current state data of the battery into the battery state estimation model to output an estimated state of the battery through the battery state estimation model, and the estimated state comprises residual capacity, a healthy state, a charge-discharge curve and temperature distribution.
The deep neural network comprises a hidden layer, an output layer and a comprehensive analysis layer arranged between the hidden layer and the output layer, wherein the comprehensive analysis layer is used for comprehensively analyzing output items of all the subnetworks in the hidden layer so as to fuse physical characteristics and working principles of the battery and enhance the accuracy of the battery state estimation model.
In an embodiment, the hybrid physical model includes at least an equivalent circuit model, a thermodynamic model and an aging model, and the estimated states correspondingly obtained include at least a state of charge (SOC), a state of health (SOH), a charge-discharge curve and a temperature distribution.
According to the battery state estimation method provided by the invention, the deep neural network is designed based on the hybrid physical model of the battery, the comprehensive analysis layer is arranged between the hidden layer and the output layer of the deep neural network, the output items of the hidden layer are comprehensively analyzed to fuse various physical characteristics and working principles of the battery, the training output results of the models are mutually verified and revised, the prediction error of a single model is reduced, and the accuracy of the finally output battery estimated state is improved. And through comprehensive analysis, the accuracy of the trained model can be improved under the condition of not increasing data samples, and the sample acquisition cost can be reduced.
The deep neural network may be selected from a multi-layer perceptron, a convolutional neural network, a cyclic neural network, a graph neural network, and the like, in a specific implementation, the convolutional neural network is selected to facilitate the subsequent extraction of the space-time characteristics of the battery, wherein the characteristic data required to extract the space-time characteristics is a part of a plurality of data provided by the battery state estimation model obtained in the embodiment, a part of the sub-networks of the deep neural network may be selected to be designed as the convolutional neural network correspondingly, each sub-network node may be designed correspondingly according to the corresponding processing requirement, and the final deep neural network may be a hybrid neural network.
The hybrid physical model of the battery can be selected from concrete battery models such as a Rint model (battery equivalent circuit), a Thevenin model, a P2D model, an SEI model and the like.
In this embodiment, the output items of the hidden layer include data, features, and models.
Wherein the data includes physical quantity and state quantity of the battery, and comprehensive analysis of the data includes normalization, standardization, dimension reduction, clustering treatment, and in this embodiment, the physical quantity and state quantity of the battery includes voltage, current, temperature, capacity, and the like of the battery.
The characteristics comprise space-time characteristics, frequency domain characteristics and statistical characteristics which characterize the internal characteristics of the battery, and the comprehensive analysis of the characteristics comprises combination, transformation, selection and noise reduction treatment.
The model comprises the equivalent circuit model, the thermodynamic model and the aging model after training, and the comprehensive analysis of the model comprises weighting, fusion, pruning and regularization treatment.
In this embodiment, before the step of training the deep neural network according to the data samples, the method further includes: and migrating the weight data in the trained similar battery state estimation model to the deep neural network as initial weight data of the deep neural network.
The similar battery and the target battery are different types of batteries or different brands of batteries, and certain similarity exists in the chemical batteries, so that the trained similar battery state estimation model is transferred to the deep neural network of the target battery, and the training period of the deep neural network of the target battery can be effectively shortened.
In this embodiment, the step of training the deep neural network according to the data samples further includes: and performing enhanced expansion on the data samples to increase the number of the data samples. To reduce data sample requirements.
The specific operation of enhancing and expanding the data sample comprises operations of rotation, scaling, clipping, noise adding and the like.
In this embodiment, the step of training the deep neural network according to the data samples further includes: and simultaneously executing a plurality of learning training tasks in the deep neural network, wherein the plurality of learning training tasks at least comprise SOC estimation training, SOH estimation training and charge-discharge curve prediction training.
The multi-task learning technology is adopted to simultaneously execute a plurality of learning training tasks, and comprehensive analysis processing is combined, so that knowledge sharing and interaction between tasks can be realized, and training speed and prediction accuracy are improved.
In this embodiment, the step of training the deep neural network according to the data samples further includes: training the deep neural network according to data samples of various batteries and a meta-learning technology to obtain a deep neural network with learning capability; training the deep neural network with learning capability according to the data sample of the target battery to obtain a battery state estimation model of the target battery.
The learning capacity of the deep neural network is expanded through a meta-learning technology, the applicability of the deep neural network to a new type of lithium ion battery can be improved, the training period of retraining can be reduced when the trained battery state estimation model is used for the new type of lithium ion battery, and the practicability of the battery state estimation model is improved.
In this embodiment, the step of training the deep neural network according to the data samples further includes: and improving the training speed of the deep neural network according to the hardware accelerator. The training period can be reduced.
The hardware acceleration can be an accelerator such as GPU, TPU, FPGA, and the corresponding software architecture can be a framework such as TensorFlow, pyTorch.
In this embodiment, the step of training the deep neural network according to the data samples further includes: and uploading training data of the deep neural network to a cloud according to cloud computing, and training the deep neural network in real time according to edge computing. Through real-time training and cloud storage, the test experiment efficiency of model training can be improved, the training test period is shortened, and the research and development period is shortened.
In this embodiment, the method further includes extracting an interpretability feature according to the battery state estimation model, and obtaining an aging influence factor of the battery according to the interpretability feature.
The method comprises the steps of analyzing the contribution degree of each input feature to an output result by locally linearly approximating a trained deep neural network by using a LIME technology, or evaluating the feature importance of the trained deep neural network based on a game theory according to a SHAP (SHapley Additive exPlanations saprolip adding and interpreting) technology.
In the battery aging analysis, the influence factors on the battery aging mainly comprise factors such as temperature, current, voltage and the like, wherein in a specific implementation, the trained battery state estimation model is subjected to local linear approximation, and the obtained battery aging influence factors can be used for optimizing the work scheduling of the battery.
In this embodiment, the interpretable feature is further provided to the comprehensive analysis layer, and the comprehensive analysis of the interpretable feature includes integrating, promoting, verifying, and correcting.
The battery state estimation method provided by the invention is based on the hybrid physical model design depth neural network of the battery, and a comprehensive analysis layer is arranged between the hidden layer and the output layer of the depth neural network, the output of the hidden layer is comprehensively analyzed to fuse various physical characteristics and working principles of the battery, the training output results of all models are mutually verified and revised, the trained battery state estimation model has high estimation accuracy on the battery state, the battery residual capacity, the health state, the charge-discharge curve and the temperature distribution can be estimated at the same time, the trained battery state estimation model comprises an equivalent circuit model, a thermodynamic model and an aging model of the battery, a battery management system of an electric automobile can conveniently obtain more comprehensive battery state information according to the various states and models, comprehensive management is carried out, and the battery management efficiency is improved.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
The foregoing examples illustrate only a few specific embodiments of the invention, which are described in greater detail and are not to be construed as limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Claims (7)
1. A battery state estimation method, characterized by comprising:
establishing a hybrid physical model of the battery according to physical characteristics and working principles of the battery to construct a deep neural network according to the hybrid physical model, wherein the hybrid physical model comprises an equivalent circuit model, a thermodynamic model and an aging model;
acquiring a data sample, and training the deep neural network according to the data sample to obtain a battery state estimation model;
acquiring current state data of a battery, wherein the current state data of the battery comprises voltage, current and temperature, and inputting the current state data of the battery into a battery state estimation model to simultaneously output an estimated state of the battery through the battery state estimation model, wherein the estimated state comprises residual capacity, a healthy state, a charge-discharge curve and temperature distribution;
extracting an interpretable feature according to the battery state estimation model, and obtaining an aging influence factor of the battery according to the interpretable feature;
the deep neural network comprises a hidden layer, an output layer and a comprehensive analysis layer arranged between the hidden layer and the output layer, wherein the comprehensive analysis layer is used for comprehensively analyzing output items of all sub-networks in the hidden layer so as to mutually verify and revise data of all models and estimated states, fuse physical characteristics and working principles of the battery and enhance the accuracy of the battery state estimation model;
the output items of the hidden layer comprise data, models and features;
the data comprise physical quantity and state quantity of the battery, and the comprehensive analysis of the data by the comprehensive analysis layer comprises normalization, standardization, dimension reduction and clustering treatment;
the model comprises the equivalent circuit model, the thermodynamic model and the aging model after training, and the comprehensive analysis of the model by the comprehensive analysis layer comprises weighting, fusion, pruning and regularization treatment;
the characteristics comprise space-time characteristics, frequency domain characteristics and statistical characteristics which characterize the internal characteristics of the battery, and the comprehensive analysis of the characteristics by the comprehensive analysis layer comprises combination, transformation, selection and noise reduction treatment;
the interpretable features are also provided to the comprehensive analysis layer, and the comprehensive analysis of the interpretable features by the comprehensive analysis layer comprises integration, popularization, verification and correction processing.
2. The battery state estimation method of claim 1, further comprising, prior to the step of training the deep neural network from the data samples:
and migrating the weight data in the trained similar battery state estimation model to the deep neural network as initial weight data of the deep neural network.
3. The battery state estimation method of claim 1, wherein training the deep neural network from the data samples further comprises:
and performing enhanced expansion on the data samples to increase the number of the data samples.
4. The battery state estimation method of claim 1, wherein training the deep neural network from the data samples further comprises:
and simultaneously executing a plurality of learning training tasks in the deep neural network, wherein the plurality of learning training tasks at least comprise SOC estimation training, SOH estimation training and charge-discharge curve prediction training.
5. The battery state estimation method of claim 1, wherein training the deep neural network from the data samples further comprises:
training the deep neural network according to data samples of various batteries and a meta-learning technology to obtain a deep neural network with learning capability;
training the deep neural network with learning capability according to the data sample of the target battery to obtain a battery state estimation model of the target battery.
6. The battery state estimation method of claim 1, wherein training the deep neural network from the data samples further comprises: and improving the training speed of the deep neural network according to the hardware accelerator.
7. The battery state estimation method of claim 1, wherein training the deep neural network from the data samples further comprises:
uploading training data of the deep neural network to a cloud according to cloud computing;
and training the deep neural network in real time according to edge calculation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310879687.XA CN116609677B (en) | 2023-07-18 | 2023-07-18 | Battery state estimation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310879687.XA CN116609677B (en) | 2023-07-18 | 2023-07-18 | Battery state estimation method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116609677A CN116609677A (en) | 2023-08-18 |
CN116609677B true CN116609677B (en) | 2023-11-03 |
Family
ID=87676758
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310879687.XA Active CN116609677B (en) | 2023-07-18 | 2023-07-18 | Battery state estimation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116609677B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117192377B (en) * | 2023-08-28 | 2024-05-07 | 广东工业大学 | Priori knowledge enhancement and semi-supervised learning-based lithium battery SOH estimation method |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114580262A (en) * | 2021-11-18 | 2022-06-03 | 吉林大学 | Lithium ion battery health state estimation method |
CN115015760A (en) * | 2022-05-10 | 2022-09-06 | 香港中文大学(深圳) | Lithium battery health state evaluation method based on neural network and migration integrated learning |
CN115792680A (en) * | 2022-12-26 | 2023-03-14 | 中山大学·深圳 | Retired battery health state assessment method based on Bayes deep learning |
CN116298914A (en) * | 2023-03-20 | 2023-06-23 | 河北工业大学 | Lithium battery state of charge and health state joint estimation method based on deep learning |
CN116381505A (en) * | 2023-04-10 | 2023-07-04 | 北京航空航天大学 | Lithium battery health state estimation method based on dynamic working condition of variation modal decomposition |
CN116413629A (en) * | 2023-03-24 | 2023-07-11 | 北京理工大学 | Spacecraft lithium battery health state estimation method based on physical information neural network |
-
2023
- 2023-07-18 CN CN202310879687.XA patent/CN116609677B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114580262A (en) * | 2021-11-18 | 2022-06-03 | 吉林大学 | Lithium ion battery health state estimation method |
CN115015760A (en) * | 2022-05-10 | 2022-09-06 | 香港中文大学(深圳) | Lithium battery health state evaluation method based on neural network and migration integrated learning |
CN115792680A (en) * | 2022-12-26 | 2023-03-14 | 中山大学·深圳 | Retired battery health state assessment method based on Bayes deep learning |
CN116298914A (en) * | 2023-03-20 | 2023-06-23 | 河北工业大学 | Lithium battery state of charge and health state joint estimation method based on deep learning |
CN116413629A (en) * | 2023-03-24 | 2023-07-11 | 北京理工大学 | Spacecraft lithium battery health state estimation method based on physical information neural network |
CN116381505A (en) * | 2023-04-10 | 2023-07-04 | 北京航空航天大学 | Lithium battery health state estimation method based on dynamic working condition of variation modal decomposition |
Also Published As
Publication number | Publication date |
---|---|
CN116609677A (en) | 2023-08-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111999657B (en) | Method for evaluating driving mileage of lithium ion battery of electric vehicle in residual life | |
Zhao et al. | A compact methodology via a recurrent neural network for accurate equivalent circuit type modeling of lithium-ion batteries | |
Landi et al. | Measurement techniques for online battery state of health estimation in vehicle-to-grid applications | |
CN116609677B (en) | Battery state estimation method | |
CN111999649A (en) | XGboost algorithm-based lithium battery residual life prediction method | |
CN110687450B (en) | Lithium battery residual life prediction method based on phase space reconstruction and particle filtering | |
CN112557907A (en) | SOC estimation method of electric vehicle lithium ion battery based on GRU-RNN | |
CN108876052A (en) | Electric car charging load forecasting method, device and computer equipment | |
AU2021101964A4 (en) | Artificial intelligence based smart electric vehicle battery management system | |
CN112666479A (en) | Battery service life prediction method based on charging cycle fusion | |
CN116859259A (en) | Battery health state estimation method based on integrated learning and real vehicle big data | |
CN115730635A (en) | Electric vehicle load prediction method | |
CN113657030A (en) | Method for predicting remaining service life of lithium battery based on Gaussian process regression | |
CN117554846B (en) | Lithium battery life prediction method and system considering constraint conditions | |
CN117825998A (en) | Lithium ion battery remaining life prediction method, system, equipment and medium | |
CN114295994A (en) | PCA-RVM-based lithium ion battery remaining service life prediction method | |
CN115684936A (en) | Power battery health failure probability calculation and health failure map establishment method based on machine learning | |
CN115700717A (en) | Power distribution analysis method based on electric automobile power consumption demand | |
CN115542168A (en) | Lithium battery residual service life prediction method based on fusion data driving model | |
CN117471320A (en) | Battery state of health estimation method and system based on charging fragments | |
CN115097312A (en) | Lithium ion battery fusion life prediction model combining data driving model and empirical model | |
CN117289167A (en) | Battery remaining life prediction method, device and medium based on multiple neural network | |
CN116930769A (en) | Lithium battery modeling method based on bidirectional generation type antagonistic neural network | |
CN117435909A (en) | Non-invasive load decomposition method based on transfer learning and multidimensional feature extraction model | |
CN117930055A (en) | Nickel-cadmium storage battery health state prediction method, device, equipment and storage medium |
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 |