CN112269134A - Battery SOC and SOH joint estimation method based on deep learning - Google Patents
Battery SOC and SOH joint estimation method based on deep learning Download PDFInfo
- Publication number
- CN112269134A CN112269134A CN202010946770.0A CN202010946770A CN112269134A CN 112269134 A CN112269134 A CN 112269134A CN 202010946770 A CN202010946770 A CN 202010946770A CN 112269134 A CN112269134 A CN 112269134A
- Authority
- CN
- China
- Prior art keywords
- network
- data
- battery
- soh
- brnn
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/385—Arrangements for measuring battery or accumulator variables
- G01R31/387—Determining ampere-hour charge capacity or SoC
- G01R31/388—Determining ampere-hour charge capacity or SoC involving voltage measurements
-
- 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]
- G01R31/3644—Constructional arrangements
- G01R31/3648—Constructional arrangements comprising digital calculation means, e.g. for performing an algorithm
-
- 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]
- G01R31/389—Measuring internal impedance, internal conductance or related variables
-
- 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
Abstract
The invention discloses a battery SOC and SOH joint estimation method based on deep learning. The method comprises the specific steps of collecting battery data, preprocessing the data, building an SE-CNN neural network and a BRNN neural network, constructing and training an estimation model, estimating the SOH value of the battery by using the trained SE-CNN network, and estimating the SOC value of the battery by using the BRNN network. The method utilizes the parameter self-learning capability of deep learning to reduce the calculated amount in the process of estimating the change relation between the ratio of the residual capacity and the maximum available capacity of the battery and the capacity and the internal resistance of the battery, considers the correlation between the SOH and the SOC, carries out combined estimation, enhances the stability of a prediction model, improves the accuracy of calculation, makes up the defects of long measurement time, high requirement on measurement conditions, overhigh calculated amount and low estimation accuracy of various estimation methods in the prior art, and provides a quick and accurate SOH and SOC estimation method for various battery management systems.
Description
Technical Field
The invention belongs to the technical field of power battery management, and particularly relates to a battery SOC and SOH joint estimation method based on deep learning.
Background
The definition of the SOC is the ratio of the residual capacity of the battery to the maximum available capacity, and the SOC reflects the residual capacity of the battery; SOH is generally defined as the relationship between the capacity and internal resistance of a lithium ion battery, and reflects the aging condition of the lithium ion battery. The accurate estimation of the SOC and the SOH is one of core functions of a battery management system, and has important significance for guaranteeing the safe use of the battery and prolonging the cycle life of the battery.
The SOC of the battery is related to a plurality of factors, such as temperature, polarization effect, battery life and the like, and has strong nonlinearity, and for the estimation of the SOC of the battery, methods generally adopted at home and abroad comprise a discharge test method, an ampere-hour method, an open-circuit voltage method, an internal resistance method, a Kalman filtering method, a linear model method and a neural network method, but the methods have defects in different degrees.
1) The discharge test method requires the battery to be in a constant current discharge state and takes a large amount of measurement time;
2) the ampere-hour method is easily influenced by the current measurement precision, and the precision is very poor under the condition of high temperature or severe current fluctuation;
3) when the open-circuit voltage method is used for estimating the SOC of the battery, the battery must be kept stand for a long time to reach a stable state, and the method is only suitable for estimating the SOC of the battery in a non-violent change state and cannot meet the requirement of online detection;
4) the internal resistance method needs to accurately measure the internal resistance of the battery, generally, the internal resistance of the battery is in the milliohm level, the requirement on a measuring instrument is very high, and the method is difficult to be applied in practice;
5) the Kalman filtering method is a more estimation method adopted at present, has stronger dependence on a battery model, needs to establish a more accurate battery model to obtain accurate SOC, and has the accuracy and the complexity which are in direct proportion;
6) the linear model is roughly classified into an equivalent circuit model and a simplified electrochemical model. The equivalent circuit model mostly adopts a simpler Thevein model, and if a more complex circuit model is adopted, the identification of model parameters and the establishment of a state equation are very challenging. Due to the simplification of the model, the internal rules of the battery cannot be fully reflected, and a large estimation error is caused.
Therefore, in the prior art, a comprehensive and effective solution for estimating the SOC and SOH of the lithium ion battery does not exist, and optimization needs to be performed in the aspects of improving accuracy and reducing calculation amount.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a battery SOC and SOH joint estimation method based on deep learning, which utilizes the parameter self-learning capability of the deep learning to reduce the calculated amount in the estimation process, considers the correlation between the SOH and the SOC to carry out joint estimation, enhances the stability of a prediction model, improves the calculation accuracy and provides a quick and accurate SOH and SOC estimation method for various battery management systems.
A battery SOC and SOH joint estimation method based on deep learning specifically comprises the following steps:
step one, battery data acquisition.
1.1, carrying out charge and discharge experiments on the lithium ion battery to obtain training data required by the SOC and SOH joint estimation model of the lithium ion battery, wherein the training data comprises voltage, current, temperature and electric quantity based on current integral of the lithium ion battery.
1.2, generating data samples by utilizing the countermeasure network to expand the sample data set. The countermeasure network consists of a generation network G and a discrimination network D. The method of generating a data sample comprises the steps of:
1.2.1, inputting random noise z into the generating network G, and generating network G output result data G (z).
1.2.2, inputting result data G (z) of a generating network G into a discrimination network D, comparing the input data G (z) with real battery data x by the discrimination network D through an objective function loss of a countermeasure network to obtain the accuracy of the result data G (z), wherein the expression of the objective function loss is as follows:
wherein the content of the first and second substances,
and
to a desired value, maximize
Means that D (x) approaches 1; minimization
Indicates that D (G (z)) is close to 0. D (x) represents an error between the real data calculated by the discrimination network D and the result data G (z) of the generated network G, and D (G (z)) represents a probability that the result data G (z) determined by the discrimination network D is real.
1.2.3, when the target function loss reaches the optimal solution, taking the data G (z) output by the generation network G as real battery data and the actually measured real battery data x to form an enhanced data set for the subsequent joint estimation of the SOC and the SOH of the battery.
And step two, preprocessing data.
Preprocessing the enhanced data set obtained in the first step through a gradient enhanced regression tree algorithm so as to obtain relatively smooth data, wherein the processing process comprises the following steps:
2.1, calculating the negative gradient of the residual error:
wherein x isiRepresenting the input vector, yiDenotes the target vector, F (x)i) Representing a model function with x mapped to y, setting an initial value F of the model function0(x)＝c，c is a constant; l (y)i,F(xi) Represents a loss function and m represents the number of iterations.
2.2 use of training set
Training the mth regression Tree Tm。
2.3 computing regression Tree TmThe weight factor of (c):
and 2.4, repeating the steps from 2.1 to 2.3M times.
2.5, updating the gradient enhancement regression tree model:
F(xi)＝F0(x)+υγmTm(xi) (4)
and upscaling is a set contraction coefficient and is used for limiting the lifting effect of each tree on the algorithm precision, so that better model performance is obtained and overfitting can be prevented.
And step three, building an SE-CNN neural network.
And 3.1, constructing the CNN neural network.
3.1.1, constructing a CNN neural network comprising an input layer, four convolutional layers, four pooling layers, a Flatten layer, a full-link layer and an output layer.
3.1.2, introduction of an inclusion structure comprising 4 basic structures, namely 1 × 1 convolution, 3 × 3 convolution, 5 × 5 convolution and 3 × 3 max pooling. 1 × 1 convolution is added before the 3 × 3 convolution and the 5 × 5 convolution structure, and 1 × 1 convolution is added after the 3 × 3 maximum pooling to reduce the parameter quantity, so that the problem of low efficiency caused by excessive calculation of the model due to excessive parameter quantity is avoided. And combining the operation results of the 4 basic structures added with the convolution on the channel.
And 3.1.3, setting the iteration times of the CNN neural network by taking the average absolute error as an objective function and adopting an RAdam adaptive optimizer to minimize the objective function.
3.2, constructing an SE module, wherein the SE module comprises the following operations:
and 3.2.1, performing compression operation, performing feature compression along the spatial dimension, and changing each two-dimensional feature channel into a real number, wherein the real number represents the global distribution of response on the feature channel.
3.2.2, performing an excitation operation to generate a weight for each eigen-channel by a parameter w, wherein the parameter w is learned to explicitly model the correlation between eigen-channels.
And 3.2.3, performing recalibration operation, taking the weight output by the excitation operation as the importance of each feature channel after feature selection, and then weighting the feature channel by channel to the previous feature through multiplication to finish recalibration on the original feature in the channel dimension.
3.3, embedding the constructed SE module into the CNN neural network to obtain the SE-CNN neural network.
And step four, building a BRNN network.
And 3.1, superposing the two RNN networks up and down to obtain the BRNN network, wherein the BRNN network comprises an input layer, a forward hidden layer, a backward hidden layer and an output layer. For each time t, the signal input to the BRNN network is simultaneously input to two RNN networks in opposite directions, and the output signal of the BRNN network is determined by the two RNN networks.
3.2, setting six weights of an input layer, a forward hidden layer, a backward hidden layer and an output layer of the BRNN network respectively, wherein the six weights are respectively as follows: input layer to forward hidden layer, backward hidden layer (w1, w3), hidden layer to hidden layer (w2, w5), forward hidden layer, backward hidden layer to output layer (w4, w 6).
And 3.3, taking the average absolute error as an objective function, minimizing the objective function by adopting an RAdam adaptive optimizer, and then setting the step length and the iteration number of the BRNN network.
And step five, constructing and training an estimation model.
And 5.1, using the SE-CNN network constructed in the third step for estimating the SOH of the lithium ion battery, setting an objective function and an optimization algorithm for training, and learning network parameters.
And 5.2, using the BRNN constructed in the fourth step for estimating the SOC of the lithium ion battery, setting an objective function and an optimization algorithm for training, and learning network parameters.
And sixthly, carrying out combined estimation on the SOC and the SOH of the lithium ion battery.
6.1, fully charging the lithium ion battery with constant current and constant voltage to obtain the total process quantity of voltage, current and temperature and the electric quantity based on current integration
And (4) inputting the SE-CNN network trained in the step 5.1, and outputting to obtain the SOH value of the lithium ion battery in the current state.
6.2, obtaining the SOH estimated value of the lithium ion battery at the time t and the SOH estimated value from the time t to the time t from the output of the 6.10Voltage, current, temperature measurement and electric quantity based on current integration over a period of time
Inputting the BRNN network trained in the step 5.2, and outputting to obtain the SOC estimated value of the lithium ion battery at the time t.
The invention has the following beneficial effects:
1. the deep learning belongs to end-to-end learning, and a result can be obtained after data is input, so that the method is convenient and quick, and the calculation speed is increased.
2. The deep learning can be used for learning the rule by optimizing the loss function without manual design of a user.
3. Potential features of data are mined, a software algorithm is improved, and estimation accuracy of the SOC is effectively improved.
Drawings
FIG. 1 is a block diagram of a SE-CNN network in the method of the present invention;
FIG. 2 is a block diagram of a BRNN network in accordance with the present invention;
FIG. 3 is a diagram of a combined SOC and SOH estimation model of a lithium ion battery;
Detailed Description
The invention is further explained below with reference to the drawings;
step one, acquiring battery data;
1.1, performing a charge-discharge experiment on the lithium ion battery to obtain training data required by a lithium ion battery SOC and SOH combined estimation model, wherein the training data comprises voltage, current, temperature and electric quantity based on current integral of the lithium ion battery;
1.2, generating data samples by utilizing a countermeasure network to expand a sample data set; the countermeasure network consists of a generation network G and a discrimination network D; the method of generating a data sample comprises the steps of:
1.2.1, inputting random noise z into a generating network G, and generating network G output result data G (z);
1.2.2, inputting result data G (z) of a generating network G into a discrimination network D, comparing the input data G (z) with real battery data x by the discrimination network D through an objective function loss of a countermeasure network to obtain the accuracy of the result data G (z), wherein the expression of the objective function loss is as follows:
wherein the content of the first and second substances,
and
to a desired value, maximize
Means that D (x) approaches 1; minimization
Represents that D (G (z)) approaches 0; d (x) represents the error between the real data obtained by the calculation of the discrimination network D and the result data G (z) of the generated network G, and D (G (z)) represents the real probability of the result data G (z) judged by the discrimination network D;
1.2.3, when the target function loss reaches the optimal solution, taking data G (z) output by the generation network G as real battery data and actually measured real battery data x to form an enhanced data set for subsequent joint estimation of the SOC and the SOH of the battery;
step two, data preprocessing;
preprocessing the enhanced data set obtained in the first step through a gradient enhanced regression tree algorithm so as to obtain relatively smooth data, wherein the processing process comprises the following steps:
2.1, calculating the negative gradient of the residual error:
wherein x isiRepresenting the input vector, yiDenotes the target vector, F (x)i) Representing a model function with x mapped to y, setting an initial value F of the model function0(x) C is a constant; l (y)i,F(xi) Represents a loss function, m represents the number of iterations;
2.2 use of training set
Training the mth regression Tree Tm；
2.3 computing regression Tree TmThe weight factor of (c):
2.4, repeating the step 2.1 to the step 2.3M times;
2.5, updating the gradient enhancement regression tree model:
F(xi)＝F0(x)+υγmTm(xi) (4)
the upscaling coefficient is a set contraction coefficient and is used for limiting the lifting effect of each tree on the algorithm precision, so that better model performance is obtained and overfitting can be prevented;
and step three, building an SE-CNN neural network.
And 3.1, constructing the CNN neural network.
3.1.1, constructing a CNN neural network comprising an input layer, four convolutional layers, four pooling layers, a Flatten layer, a full-link layer and an output layer.
3.1.2, introduction of an inclusion structure comprising 4 basic structures, namely 1 × 1 convolution, 3 × 3 convolution, 5 × 5 convolution and 3 × 3 max pooling; 1, 1 convolution is added before the 3 × 3 convolution and the 5 × 5 convolution structure, and 1 convolution 1 is added after the 3 × 3 maximum pooling to reduce the parameter quantity, so that the problem of low efficiency caused by overlarge calculated quantity of the model due to overlarge parameter quantity is avoided; combining the operation results of the 4 basic structures added with the convolution on the channel;
3.1.3, setting the iteration times of the CNN neural network by taking the average absolute error as a target function and adopting an RAdam adaptive optimizer to minimize the target function;
3.2, constructing an SE module, wherein the SE module comprises the following operations:
and 3.2.1, performing compression operation, performing feature compression along the spatial dimension, and changing each two-dimensional feature channel into a real number, wherein the real number represents the global distribution of response on the feature channel.
3.2.2, performing an excitation operation to generate a weight for each eigen-channel by a parameter w, wherein the parameter w is learned to explicitly model the correlation between eigen-channels.
And 3.2.3, performing recalibration operation, taking the weight output by the excitation operation as the importance of each feature channel after feature selection, and then weighting the feature channel by channel to the previous feature through multiplication to finish recalibration on the original feature in the channel dimension.
3.3, embedding the constructed SE module into the CNN neural network to obtain the SE-CNN neural network, as shown in FIG. 1.
And step four, building a BRNN network.
And 3.1, superposing the two RNN networks up and down to obtain the BRNN, wherein the BRNN comprises an input layer, a forward hidden layer, a backward hidden layer and an output layer, as shown in figure 2. For each time t, the signal input to the BRNN network is simultaneously input to two RNN networks in opposite directions, and the output signal of the BRNN network is determined by the two RNN networks.
3.2, setting six weights of an input layer, a forward hidden layer, a backward hidden layer and an output layer of the BRNN network respectively, wherein the six weights are respectively as follows: input layer to forward hidden layer, backward hidden layer (w1, w3), hidden layer to hidden layer (w2, w5), forward hidden layer, backward hidden layer to output layer (w4, w 6).
And 3.3, taking the average absolute error as an objective function, minimizing the objective function by adopting an RAdam adaptive optimizer, and then setting the step length and the iteration number of the BRNN network.
And step five, constructing and training an estimation model.
And 5.1, using the SE-CNN network constructed in the third step for estimating the SOH of the lithium ion battery, setting an objective function and an optimization algorithm for training, and learning network parameters.
And 5.2, using the BRNN constructed in the fourth step for estimating the SOC of the lithium ion battery, setting an objective function and an optimization algorithm for training, and learning network parameters.
And sixthly, performing combined estimation on the SOC and the SOH of the lithium ion battery as shown in the figure 3.
6.1, fully charging the lithium ion battery with constant current and constant voltage to obtain the total process quantity of voltage, current and temperature and the electric quantity based on current integration
And (4) inputting the SE-CNN network trained in the step 5.1, and outputting to obtain the SOH value of the lithium ion battery in the current state.
6.2, obtaining the SOH estimated value of the lithium ion battery at the time t and the SOH estimated value from the time t to the time t from the output of the 6.10Voltage, current, temperature measurement and electric quantity based on current integration over a period of time
And inputting the BRNN network trained in the step 5.2, and outputting to obtain the SOC estimated value of the lithium ion battery at the time t, wherein compared with the prior art, the SOC estimated precision of the lithium ion battery obtained by the embodiment is improved by 5%.
Claims (1)
1. A battery SOC and SOH joint estimation method based on deep learning is characterized in that: the method specifically comprises the following steps:
step one, acquiring battery data;
1.1, performing a charge-discharge experiment on the lithium ion battery to obtain training data required by a lithium ion battery SOC and SOH combined estimation model, wherein the training data comprises voltage, current, temperature and electric quantity based on current integral of the lithium ion battery;
1.2, generating data samples by utilizing a countermeasure network to expand a sample data set; the countermeasure network consists of a generation network G and a discrimination network D; the method of generating a data sample comprises the steps of:
1.2.1, inputting random noise z into a generating network G, and generating network G output result data G (z);
1.2.2, inputting result data G (z) of a generating network G into a discrimination network D, comparing the input data G (z) with real battery data x by the discrimination network D through an objective function loss of a countermeasure network to obtain the accuracy of the result data G (z), wherein the expression of the objective function loss is as follows:
wherein the content of the first and second substances,
and
to a desired value, maximize
Represents D (x) approachAt 1; minimization
Represents that D (G (z)) approaches 0; d (x) represents the error between the real data obtained by the calculation of the discrimination network D and the result data G (z) of the generated network G, and D (G (z)) represents the real probability of the result data G (z) judged by the discrimination network D;
1.2.3, when the target function loss reaches the optimal solution, taking data G (z) output by a generating network G as real battery data and actually measured real battery data x to form an enhanced data set;
step two, data preprocessing;
preprocessing the enhanced data set obtained in the first step through a gradient enhanced regression tree algorithm, wherein the processing process comprises the following steps:
2.1, calculating the negative gradient of the residual error:
wherein x isiRepresenting the input vector, yiDenotes the target vector, F (x)i) Representing a model function with x mapped to y, setting an initial value F of the model function0(x) C is a constant; l (y)i,F(xi) Represents a loss function, m represents the number of iterations;
2.2 use of training set
Training the mth regression Tree Tm；
2.3 computing regression Tree TmThe weight factor of (c):
2.4, repeating the step 2.1 to the step 2.3M times;
2.5, updating the gradient enhancement regression tree model:
F(xi)＝F0(x)+υγmTm(xi) (4)
wherein upsilon is a set shrinkage factor;
step three, building an SE-CNN neural network;
3.1, constructing a CNN neural network;
3.1.1, constructing a CNN neural network comprising an input layer, four convolutional layers, four pooling layers, a Flatten layer, a full-connection layer and an output layer;
3.1.2, introduction of an inclusion structure comprising 4 basic structures, namely 1 × 1 convolution, 3 × 3 convolution, 5 × 5 convolution and 3 × 3 max pooling; adding 1 convolution 1 x 1 before the structure of 3 x 3 convolution and 5 x 5 convolution, adding 1 convolution 1 x 1 after 3 x 3 max pooling to reduce the parameter; combining the operation results of the 4 basic structures added with the convolution on the channel;
3.1.3, setting the iteration times of the CNN neural network by taking the average absolute error as a target function and adopting an RAdam adaptive optimizer to minimize the target function;
3.2, constructing an SE module, wherein the SE module comprises the following operations:
3.2.1, performing compression operation, performing feature compression along the spatial dimension, and changing each two-dimensional feature channel into a real number, wherein the real number represents the global distribution of response on the feature channel;
3.2.2, performing an excitation operation, and generating a weight for each characteristic channel through a parameter w, wherein the parameter w is learned to be used for explicitly modeling the correlation among the characteristic channels;
3.2.3, performing recalibration operation, taking the weight output by the excitation operation as the importance of each feature channel after feature selection, and then weighting the feature channel by channel through multiplication to the previous feature to finish recalibration on the original feature in the channel dimension;
3.3, embedding the constructed SE module into the CNN neural network to obtain an SE-CNN neural network;
step four, building a BRNN network;
3.1, superposing the two RNN networks up and down to obtain a BRNN network, wherein the BRNN network comprises an input layer, a forward hidden layer, a backward hidden layer and an output layer; for each time t, the signal input into the BRNN network is simultaneously input into two RNN networks with opposite directions, and the output signal of the BRNN network is determined by the two RNN networks;
3.2, setting six weights of an input layer, a forward hidden layer, a backward hidden layer and an output layer of the BRNN network respectively, wherein the six weights are respectively as follows: input layer to forward hidden layer, backward hidden layer (w1, w3), hidden layer to hidden layer (w2, w5), forward hidden layer, backward hidden layer to output layer (w4, w 6);
3.3, taking the average absolute error as a target function, adopting an RAdam adaptive optimizer to minimize the target function, and then setting the step length and the iteration times of the BRNN network;
constructing and training an estimation model;
5.1, the SE-CNN network constructed in the third step is used for estimating the SOH of the lithium ion battery, an objective function and an optimization algorithm are set for training, and network parameters are learned;
5.2, the BRNN network constructed in the fourth step is used for estimating the SOC of the lithium ion battery, an objective function and an optimization algorithm are set for training, and network parameters are learned;
step six, carrying out combined estimation on the SOC and the SOH of the lithium ion battery;
6.1, fully charging the lithium ion battery with constant current and constant voltage to obtain the total process quantity of voltage, current and temperature and the electric quantity based on current integration
Inputting the SE-CNN network trained in the step 5.1, and outputting to obtain the SOH value of the lithium ion battery in the current state;
6.2, obtaining the SOH estimated value of the lithium ion battery at the time t and the SOH estimated value from the time t to the time t from the output of the 6.10Voltage, current, temperature measurement and electric quantity based on current integration over a period of time
Inputting the BRNN network trained in the step 5.2, and outputting to obtain the SOC estimated value of the lithium ion battery at the time t.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010946770.0A CN112269134B (en) | 2020-09-10 | 2020-09-10 | Battery SOC and SOH joint estimation method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010946770.0A CN112269134B (en) | 2020-09-10 | 2020-09-10 | Battery SOC and SOH joint estimation method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112269134A true CN112269134A (en) | 2021-01-26 |
| CN112269134B CN112269134B (en) | 2022-12-02 |
Family
ID=74349483
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010946770.0A Active CN112269134B (en) | 2020-09-10 | 2020-09-10 | Battery SOC and SOH joint estimation method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112269134B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113093014A (en) * | 2021-03-31 | 2021-07-09 | 山东建筑大学 | Online collaborative estimation method and system for SOH and SOC based on impedance parameters |
| CN113391225A (en) * | 2021-05-19 | 2021-09-14 | 北京航空航天大学 | Lithium battery state-of-charge estimation method considering capacity degradation |
| CN113740739A (en) * | 2021-07-22 | 2021-12-03 | 西安交通大学 | Method for predicting residual life of vehicle-mounted starting maintenance-free lead-acid storage battery |
| CN113884905A (en) * | 2021-11-02 | 2022-01-04 | 山东大学 | Power battery state of charge estimation method and system based on deep learning |
| CN114114052A (en) * | 2021-11-25 | 2022-03-01 | 福州大学 | Method for rapidly estimating SOH and SOC of battery based on shock response characteristic |
| CN114236410A (en) * | 2021-11-24 | 2022-03-25 | 广东工业大学 | Lithium battery SOH estimation method based on novel generation type antagonistic neural network |
| CN117148161A (en) * | 2023-08-29 | 2023-12-01 | 深圳市今朝时代股份有限公司 | Battery SOC estimation method and device based on cloud neural network |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105353313A (en) * | 2015-09-28 | 2016-02-24 | 欣旺达电子股份有限公司 | Estimation method of battery state of charge and estimation device of battery state of charge |
| CN106918789A (en) * | 2017-05-10 | 2017-07-04 | 成都理工大学 | A kind of SOC SOH combine online real-time estimation and on-line amending method |
| CN107037374A (en) * | 2017-05-04 | 2017-08-11 | 广东电网有限责任公司电力科学研究院 | A kind of SOC and SOH combined estimation methods of lithium ion battery |
| US20170350944A1 (en) * | 2016-06-06 | 2017-12-07 | Mitsubishi Electric Research Laboratories, Inc. | Methods and Systems for Data-Driven Battery State of Charge (SoC) Estimation |
| JP2017223537A (en) * | 2016-06-15 | 2017-12-21 | 本田技研工業株式会社 | Device for estimating battery state and method for estimating battery state |
| CN108520986A (en) * | 2018-05-10 | 2018-09-11 | 杭州电子科技大学 | A kind of power battery method for group matching based on generation confrontation network |
| CN109802430A (en) * | 2018-12-29 | 2019-05-24 | 上海电力学院 | A kind of wind-powered electricity generation power grid control method based on LSTM-Attention network |
| CN110554324A (en) * | 2019-08-26 | 2019-12-10 | 华中科技大学 | SOC and SOH joint estimation method |
| CN110658459A (en) * | 2019-09-12 | 2020-01-07 | 北京航空航天大学 | Lithium ion battery state of charge estimation method based on bidirectional cyclic neural network |
| CN110888058A (en) * | 2019-12-02 | 2020-03-17 | 西安科技大学 | Algorithm based on power battery SOC and SOH joint estimation |
| CN111007399A (en) * | 2019-11-15 | 2020-04-14 | 浙江大学 | Lithium battery state of charge prediction method based on improved generation countermeasure network |
| CN111044928A (en) * | 2019-12-31 | 2020-04-21 | 福州大学 | Lithium battery health state estimation method |
-
2020
- 2020-09-10 CN CN202010946770.0A patent/CN112269134B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105353313A (en) * | 2015-09-28 | 2016-02-24 | 欣旺达电子股份有限公司 | Estimation method of battery state of charge and estimation device of battery state of charge |
| US20170350944A1 (en) * | 2016-06-06 | 2017-12-07 | Mitsubishi Electric Research Laboratories, Inc. | Methods and Systems for Data-Driven Battery State of Charge (SoC) Estimation |
| JP2017223537A (en) * | 2016-06-15 | 2017-12-21 | 本田技研工業株式会社 | Device for estimating battery state and method for estimating battery state |
| CN107037374A (en) * | 2017-05-04 | 2017-08-11 | 广东电网有限责任公司电力科学研究院 | A kind of SOC and SOH combined estimation methods of lithium ion battery |
| CN106918789A (en) * | 2017-05-10 | 2017-07-04 | 成都理工大学 | A kind of SOC SOH combine online real-time estimation and on-line amending method |
| CN108520986A (en) * | 2018-05-10 | 2018-09-11 | 杭州电子科技大学 | A kind of power battery method for group matching based on generation confrontation network |
| CN109802430A (en) * | 2018-12-29 | 2019-05-24 | 上海电力学院 | A kind of wind-powered electricity generation power grid control method based on LSTM-Attention network |
| CN110554324A (en) * | 2019-08-26 | 2019-12-10 | 华中科技大学 | SOC and SOH joint estimation method |
| CN110658459A (en) * | 2019-09-12 | 2020-01-07 | 北京航空航天大学 | Lithium ion battery state of charge estimation method based on bidirectional cyclic neural network |
| CN111007399A (en) * | 2019-11-15 | 2020-04-14 | 浙江大学 | Lithium battery state of charge prediction method based on improved generation countermeasure network |
| CN110888058A (en) * | 2019-12-02 | 2020-03-17 | 西安科技大学 | Algorithm based on power battery SOC and SOH joint estimation |
| CN111044928A (en) * | 2019-12-31 | 2020-04-21 | 福州大学 | Lithium battery health state estimation method |
Non-Patent Citations (1)
| Title |
|---|
| 李超然 等: "基于深度学习的锂离子电池SOC和SOH联合估算", 《中国电机工程学报》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113093014A (en) * | 2021-03-31 | 2021-07-09 | 山东建筑大学 | Online collaborative estimation method and system for SOH and SOC based on impedance parameters |
| CN113093014B (en) * | 2021-03-31 | 2022-05-27 | 山东建筑大学 | Online collaborative estimation method and system for SOH and SOC based on impedance parameters |
| CN113391225A (en) * | 2021-05-19 | 2021-09-14 | 北京航空航天大学 | Lithium battery state-of-charge estimation method considering capacity degradation |
| CN113391225B (en) * | 2021-05-19 | 2022-06-03 | 北京航空航天大学 | Lithium battery state-of-charge estimation method considering capacity degradation |
| CN113740739A (en) * | 2021-07-22 | 2021-12-03 | 西安交通大学 | Method for predicting residual life of vehicle-mounted starting maintenance-free lead-acid storage battery |
| CN113740739B (en) * | 2021-07-22 | 2023-05-02 | 西安交通大学 | Vehicle-mounted starting maintenance-free lead-acid storage battery residual life prediction method |
| CN113884905A (en) * | 2021-11-02 | 2022-01-04 | 山东大学 | Power battery state of charge estimation method and system based on deep learning |
| CN113884905B (en) * | 2021-11-02 | 2022-06-14 | 山东大学 | Power battery state of charge estimation method and system based on deep learning |
| CN114236410A (en) * | 2021-11-24 | 2022-03-25 | 广东工业大学 | Lithium battery SOH estimation method based on novel generation type antagonistic neural network |
| CN114114052A (en) * | 2021-11-25 | 2022-03-01 | 福州大学 | Method for rapidly estimating SOH and SOC of battery based on shock response characteristic |
| CN117148161A (en) * | 2023-08-29 | 2023-12-01 | 深圳市今朝时代股份有限公司 | Battery SOC estimation method and device based on cloud neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112269134B (en) | 2022-12-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112269134B (en) | Battery SOC and SOH joint estimation method based on deep learning | |
| CN111832220A (en) | Lithium ion battery health state estimation method based on codec model | |
| CN113064093B (en) | Method and system for jointly estimating state of charge and state of health of energy storage battery | |
| CN111709186B (en) | Integrated estimation method for health state of retired power lithium battery | |
| CN110824363B (en) | Lithium battery SOC and SOE joint estimation method based on improved CKF | |
| CN112630659A (en) | Lithium battery SOC estimation method based on improved BP-EKF algorithm | |
| CN110082682A (en) | A kind of lithium battery charge state estimation method | |
| CN112784140B (en) | Search method of high-energy-efficiency neural network architecture | |
| CN112834927A (en) | Lithium battery residual life prediction method, system, device and medium | |
| CN114660464A (en) | Lithium ion battery state of charge estimation method | |
| Yao et al. | Novel lithium-ion battery state-of-health estimation method using a genetic programming model | |
| CN109633449A (en) | Mining service life of lithium battery prediction technique and management system based on grey vector machine | |
| CN114217234B (en) | IDE-ASRCKF-based lithium ion battery parameter identification and SOC estimation method | |
| CN110232432B (en) | Lithium battery pack SOC prediction method based on artificial life model | |
| Li et al. | State of charge estimation of lithium-ion batteries based on PSO-TCN-Attention neural network | |
| Xu et al. | A hybrid method for lithium-ion batteries state-of-charge estimation based on gated recurrent unit neural network and an adaptive unscented Kalman filter | |
| CN113033104A (en) | Lithium battery state of charge estimation method based on graph convolution | |
| CN112014757A (en) | Battery SOH estimation method integrating capacity increment analysis and genetic wavelet neural network | |
| CN109738808B (en) | Method for estimating SOC (State of Charge) based on BP (Back propagation) neural network optimized by simulated annealing algorithm | |
| CN114280490B (en) | Lithium ion battery state of charge estimation method and system | |
| CN115481796A (en) | Method for predicting remaining service life of battery based on Bayesian hybrid neural network | |
| Li et al. | Research on SOC estimation of lithium battery based on GWO-BP neural network | |
| CN113466710B (en) | SOC and SOH collaborative estimation method for energy storage battery in receiving-end power grid containing new energy | |
| CN115980588A (en) | Lithium ion battery health state estimation method based on self-encoder extraction features | |
| Zhou et al. | Hyperparameter optimization for SOC estimation by LSTM with internal resistance |
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 |