CN114779103A - Lithium ion battery SOC estimation method based on time-lag convolutional neural network - Google Patents

Lithium ion battery SOC estimation method based on time-lag convolutional neural network Download PDF

Info

Publication number
CN114779103A
CN114779103A CN202210453820.0A CN202210453820A CN114779103A CN 114779103 A CN114779103 A CN 114779103A CN 202210453820 A CN202210453820 A CN 202210453820A CN 114779103 A CN114779103 A CN 114779103A
Authority
CN
China
Prior art keywords
data
time
lag
neural network
lithium ion
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
Application number
CN202210453820.0A
Other languages
Chinese (zh)
Inventor
李俊红
蒋泽宇
顾菊平
华亮
宗天成
褚云琨
芮佳丽
张泓睿
严俊
肖康
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nantong University
Original Assignee
Nantong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nantong University filed Critical Nantong University
Priority to CN202210453820.0A priority Critical patent/CN114779103A/en
Publication of CN114779103A publication Critical patent/CN114779103A/en
Pending legal-status Critical Current

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
    • 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
    • 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

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (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 lithium ion battery SOC estimation method based on a time-lag convolutional neural network, and belongs to the technical field of lithium ion batteries. The technical problem that the time lag and the convolutional neural network cannot be directly combined is solved. The technical scheme is as follows: the method comprises the following steps: step 1) measuring the terminal voltage, current and temperature of a lithium ion battery with the SOC from 1 to 0; step 2) preprocessing the measured data, and constructing a time-lag convolutional neural network training data set and a test data set for SOC estimation; and 3) training and testing the data set by using a time-lag convolutional neural network to realize SOC real-time estimation. The beneficial effects of the invention are as follows: the time-lag convolutional neural network can consider more data and has higher estimation precision.

Description

Lithium ion battery SOC estimation method based on time-lag convolution neural network
Technical Field
The invention relates to the technical field of lithium ion batteries, in particular to a lithium ion battery SOC estimation method based on a time-lag convolutional neural network.
Background
Accurate estimation of the state of charge (SOC) of a lithium ion battery is important because it can tell the user the amount of "fuel" remaining in battery powered systems or devices, such as electric cars, drones, and cell phones.
Battery models can be divided into three major categories: an electrochemical model, an equivalent circuit model, and a machine learning model. Electrochemical models are generally considered to be computationally expensive and, therefore, are not suitable for estimating lithium ion battery SOC in real time. Although the electrochemical model can build various reduced order models by simplification and assumption, the SOC estimation accuracy is similar to the equivalent circuit model. Machine learning models have recently become more prevalent in lithium ion battery SOC estimation, with typical models including gaussian processes, feed-forward neural networks, time-lag neural networks, recurrent neural networks, convolutional neural networks, and long-and-short-term memory neural networks. The time-lag recurrent neural network successfully applies the concept of time lag to the recurrent neural network, the time-lag recurrent neural network is explained in detail in the book 'time-lag recurrent neural network', a plurality of important problems are introduced, and students use the network to research the SOC estimation of the lithium ion battery. Unlike a recurrent neural network, the invention of a convolutional neural network is intended for image processing, and an image is an input that can be regarded as a two-dimensional structure, and therefore, it is not possible to directly combine a time lag with a convolutional neural network as in a time lag recurrent neural network. Combining the time-lag concept with the convolutional neural network and using it for estimating the SOC of the lithium ion battery becomes a technical difficulty.
How to solve the above technical problems is the subject of the present invention.
Disclosure of Invention
The invention aims to provide a lithium ion battery SOC estimation method based on a time-lag convolutional neural network, which is used for acquiring various parameters of a battery through battery charging and discharging experiments at different temperatures, processing data and constructing a training data set, importing the training data set into the time-lag convolutional neural network for training, and finally estimating the SOC of the lithium ion battery in real time.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that: a lithium ion battery SOC estimation method based on a time-lag convolutional neural network comprises the following steps:
step 1) fully charging a brand-new lithium ion battery at a changing temperature, and repeatedly measuring the terminal voltage, the current and the temperature of the lithium ion battery with the SOC from 1 to 0 through a constant current discharge experiment, a DST working condition discharge experiment, an FUDS working condition discharge experiment and an US06 working condition discharge experiment;
step 2) preprocessing the measured data, and constructing a time-lag convolutional neural network training data set and a test data set for SOC estimation;
and 3) training and testing the data set by using a time-lag convolutional neural network to realize the final SOC real-time estimation.
As a further optimization scheme of the lithium ion battery SOC estimation method based on the time lag convolutional neural network provided by the invention, the step 2) specifically comprises the following steps:
the data in the step 2-1) has no unified dimension, and the problems of slow convergence speed, high error and the like exist when a neural network algorithm is fitted, so that the measured data needs to be subjected to normalization processing. The battery objects of the experiments under various working conditions are the same, but the variation ranges of voltage, current and the like are different, so that the data under various working conditions need to be preprocessed separately, and the normalization processing is carried out on the input data by adopting a maximum and minimum normalization method. The output is the SOC of the lithium ion battery, which ranges from 0 to 1, so no normalization process is required.
After the data normalization processing in the step 2-2), a data set is constructed by using the following method, firstly, the measured value in one discharge period is intercepted into a plurality of data with the length of n, and the method is as follows:
Figure BDA0003617995120000021
wherein x isijJ data, x, representing the ith cycle in the input data setijCorresponding output ti_jThe SOC value of the lithium ion battery at the moment. U represents the voltage across the battery, I represents the current passing through the battery, T represents the temperature at which the battery operates, and T represents the time of discharge. In the present invention, n is 10.
And 2-3) performing discharge experiments on the working conditions of one total cycle charge-discharge frequency N and one single discharge period m. In order to fully discover the relation between the front input data and the rear input data in each discharge period, the padding method is used for expanding the edge data, then N × m groups of input data with the size of 4 × 10 are generated in total, corresponding N × m SOC values are output data, and the input data and the output data under different working conditions are combined into a total data set.
And 2-4) in order not to influence the training of the data set, the training and testing data set is extracted without carrying out scrambling operation on the total data set. And selecting 80% of data in the front of a discharge working condition as a training data set, and selecting the 20% of data in the back as a test data set, wherein the training data set is used as training and test data of the time-lag convolution neural network.
As a further optimization scheme of the lithium ion battery SOC estimation method based on the time-lag convolutional neural network provided by the invention, the step 3) specifically comprises the following steps:
step 3-1) the time-lag convolution neural network architecture developed aiming at the lithium ion battery SOC estimation comprises the following steps: an Input layer (Input), three convolutional layers c (volume), a fully connected layer f (full connected layer), and an Output layer (Output);
step 3-2) time-lag operation of the time-lag convolutional neural network is embodied in a convolutional layer, and input data isxijIn the convolution kernel convolution operation of each convolution layer, the result is convolved with the current ith cycle data and the (i-1) th cycle xi-1jThe weighted sum of the convolution results of the data is output as the final convolution layer, and the convolution kernels of the two convolution operations have different weights, so that the time lag effect can be realized.
And 3-3) activating the convolution layer and the full-link layer by using a relu function, activating the output layer by using a tanh function, and mapping the estimated value to 0-1.
Compared with the prior art, the invention has the following beneficial effects:
(1) the time-lag concept is combined with the convolutional neural network, a time-lag convolutional neural network algorithm is provided, and the time-lag convolutional neural network algorithm can be applied to SOC estimation of the lithium ion battery.
(2) The time-lag convolutional neural network algorithm used by the invention can consider the data of the previous cycle through time-lag operation while considering the data of the past moment of the current cycle period, and can have higher estimation precision.
(3) The convolution operation of the time-lag convolution neural network in the convolution layer is to take the weighted sum of two data processed by two convolution kernels as the final convolution layer output, and the convolution neural network with a plurality of convolution kernels uses a plurality of convolution kernels to process the same data, so that the time-lag convolution neural network can consider more data and has higher estimation precision.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.
FIG. 1 is a time-lapse convolutional neural network framework of the present invention.
FIG. 2 is a schematic diagram of a time-lag convolution operation according to the present invention.
Fig. 3 is a SOC estimation result of a certain whole discharge cycle in the constant current discharge experiment test set of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. Of course, the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example 1
Referring to fig. 1 to fig. 3, the present invention provides a technical solution that the present invention provides a lithium ion battery SOC estimation method based on a time lag convolutional neural network, in this embodiment, research is performed with an under-the-pine lithium ion battery NCR-18650B as an object, a calibration voltage is 3.7V, and a battery capacity is 3400 mAh; the battery is fully charged in a constant-current and constant-voltage charging mode, after standing for 1h, the battery is in a full-charge state, the battery is subjected to discharge experiments respectively under the constant-current discharge, DST working condition, FUDS working condition and US06 working condition until the voltage is reduced to a discharge cut-off voltage, and the experiments are repeated.
In order to better achieve the purpose of the present invention, this embodiment is a lithium ion battery SOC estimation method based on a time lag convolutional neural network, which specifically includes the following steps:
step 1) fully charging a brand-new lithium ion battery at a changing temperature, and repeatedly measuring the terminal voltage, the current and the temperature of the lithium ion battery with the SOC from 1 to 0 through a constant current discharge experiment, a DST working condition discharge experiment, an FUDS working condition discharge experiment and an US06 working condition discharge experiment;
step 2) preprocessing the measured data, and constructing a time-lag convolutional neural network training data set and a test data set for SOC estimation;
and 3) training and testing the data set by using a time-lag convolutional neural network to realize the final SOC real-time estimation.
Preferably, the step 2) specifically comprises the following steps:
step 2-1), the problems of slow convergence speed, high error and the like of data without a unified dimension during the fitting of a neural network algorithm are solved, so that the measured data needs to be subjected to normalization processing; the battery objects of experiments under various working conditions are the same, but the variation ranges of voltage, current and the like are different, so that the data under various working conditions need to be preprocessed independently, the input data is normalized by adopting a maximum and minimum normalization method, the output is the SOC of the lithium ion battery, and the SOC ranges from 0 to 1, so that the normalization processing is not needed.
After the data normalization processing in the step 2-2), a data set is constructed by using the following method, firstly, the measured value in one discharge period is intercepted into a plurality of data with the length of n, and the method is as follows:
Figure BDA0003617995120000041
wherein x isijJ data, x, representing the ith cycle in the input data setijCorresponding output ti_jThe SOC value of the lithium ion battery at the moment, U represents the voltage at two ends of the battery, I represents the current passing by the battery, T represents the temperature when the battery works, and T represents the discharging time, wherein n is 10.
Step 2-3) for a working condition discharge experiment with a total cycle charge-discharge frequency of N and a single discharge period of m, in order to fully discover the relation between front and rear input data in each discharge period, edge data is expanded by using a padding method, then N x m groups of input data with the size of 4 x 10 are generated in total, corresponding N x m SOC values are output data, and input and output data of different working conditions are combined into a total data set;
and 2-4) in order not to influence the training of the data set, performing disorganization operation on the total data set, then extracting the training and testing data set, selecting 80% of data in the front of a discharge working condition as the training data set, and selecting the 20% of data in the back as the testing data set, and using the data as the training and testing data of the time-lag convolution neural network.
Preferably, the step 3) specifically comprises the following steps:
step 3-1) the time-lag convolutional neural network architecture developed for lithium ion battery SOC estimation, as shown in fig. 1, includes: an Input layer (Input), three convolution layers c (conversion), a fully connected layer f (full connected layer) and an Output layer (Output);
step 3-2) time-lag operation of the time-lag convolutional neural network is embodied in a convolutional layer, and x is taken as input dataijIn the convolution kernel convolution operation of each convolution layer, the convolution result and the (i-1) th cycle x are convolved by the current ith cycle datai-1jThe weighted sum of the convolution results of the data is output as the final convolutional layer, and the convolution kernels of the two convolution operations have different weights, as shown in fig. 2, so that a time lag effect can be achieved.
And 3-3) activating all the convolution layer and the full-connection layer by using a relu function, activating the output layer by using a tanh function, mapping the estimated value to 0-1, and after training, setting the error evaluation index MAPE of the final network to be 1.74%, wherein fig. 3 shows the SOC estimation result of a certain whole discharge period in the test set of the constant-current discharge experiment.
The time-lag concept is combined with the convolutional neural network, a time-lag convolutional neural network algorithm is provided, and the time-lag convolutional neural network algorithm can be applied to SOC estimation of the lithium ion battery.
The time-lag convolutional neural network algorithm used by the invention can consider the data of the previous cycle through time-lag operation while considering the data of the past moment of the current cycle period, and can have higher estimation precision.
The convolution operation of the time-lag convolutional neural network in the convolutional layer is to take the weighted sum of two data processed by two convolution kernels as the final convolutional layer output, and the convolutional neural network with a plurality of convolution kernels uses a plurality of convolution kernels to process the same data, so that the time-lag convolutional neural network can consider more data and has higher estimation precision.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (3)

1. A lithium ion battery SOC estimation method based on a time-lag convolutional neural network is characterized by comprising the following steps:
step 1) fully charging a brand-new lithium ion battery at a changing temperature, and repeatedly measuring the terminal voltage, the current and the temperature of the lithium ion battery with the SOC from 1 to 0 through a constant current discharge experiment, a DST working condition discharge experiment, an FUDS working condition discharge experiment and an US06 working condition discharge experiment;
step 2) preprocessing the measured data, and constructing a time-lag convolutional neural network training data set and a test data set for SOC estimation;
and 3) training and testing the data set by using a time-lag convolutional neural network to realize the final SOC real-time estimation.
2. The lithium ion battery SOC estimation method based on the time-lag convolutional neural network of claim 1, wherein the step 2) specifically comprises the following steps:
step 2-1) carrying out normalization processing on the measured data, wherein the battery objects of experiments under various working conditions are the same, the voltage and current variation ranges are different, the data under various working conditions are preprocessed independently, the input data are normalized by adopting a maximum and minimum normalization method, the output is the SOC of the lithium ion battery, the SOC range is between 0 and 1, and the normalization processing is not needed;
after the data normalization processing in the step 2-2), a data set is constructed by using the following method, firstly, the measured value in one discharge period is intercepted into a plurality of data with the length of n, and the method is as follows:
Figure FDA0003617995110000011
wherein x isijJ data, x, representing the ith cycle in the input data setijCorresponding output ti_jThe SOC value of the lithium ion battery at the moment, U represents the voltage at two ends of the battery, I represents the current passed by the battery, T represents the temperature of the battery during working, and T represents the time of discharging, wherein n is 10;
step 2-3) for a working condition discharge experiment with a total cycle charge-discharge frequency N and a single discharge period m, in order to fully discover the relation between front and rear input data in each discharge period, edge data is expanded by using a padding method, then N × m groups of input data with the size of 4 × 10 are generated in total, corresponding N × m SOC values are output data, and input and output data of different working conditions are combined into a total data set;
and 2-4) in order not to influence the training of the data set and not to disturb the total data set, extracting the training and testing data set, selecting 80% of data before a discharge working condition as the training data set and 20% of data after the discharge working condition as the testing data set, and using the data as the training and testing data of the time-lag convolution neural network.
3. The time-lag convolutional neural network-based lithium ion battery SOC estimation method of claim 1, wherein the step 3) specifically comprises the following steps:
step 3-1) the time lag convolution neural network architecture developed aiming at the lithium ion battery SOC estimation comprises the following steps: an Input layer (Input), three convolution layers c (conversion), a fully connected layer f (full connected layer) and an Output layer (Output);
step 3-2) time-lag operation of the time-lag convolutional neural network is embodied in a convolutional layer, and x is taken as input dataijIn the convolution kernel convolution operation of each convolution layer, the convolution result and the (i-1) th cycle x are convolved by the current ith cycle datai-1jThe weighted sum of the convolution results of the data is output as a final convolution layer, and convolution kernels of the two convolution operations have different weights, so that a time lag effect is realized;
and 3-3) activating all the convolution layers and the full connection layers by using a relu function, activating the output layer by using a tanh function, and mapping the estimated value to 0-1.
CN202210453820.0A 2022-04-24 2022-04-24 Lithium ion battery SOC estimation method based on time-lag convolutional neural network Pending CN114779103A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210453820.0A CN114779103A (en) 2022-04-24 2022-04-24 Lithium ion battery SOC estimation method based on time-lag convolutional neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210453820.0A CN114779103A (en) 2022-04-24 2022-04-24 Lithium ion battery SOC estimation method based on time-lag convolutional neural network

Publications (1)

Publication Number Publication Date
CN114779103A true CN114779103A (en) 2022-07-22

Family

ID=82432266

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210453820.0A Pending CN114779103A (en) 2022-04-24 2022-04-24 Lithium ion battery SOC estimation method based on time-lag convolutional neural network

Country Status (1)

Country Link
CN (1) CN114779103A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115343625A (en) * 2022-10-17 2022-11-15 力高(山东)新能源技术股份有限公司 Power battery SOC estimation method based on error correction
CN116300410A (en) * 2023-05-25 2023-06-23 武汉理工大学 Corner optimization method and system for data-driven feedforward and feedback compensation
CN117658208A (en) * 2023-12-20 2024-03-08 潮州市丰业新材料有限公司 High-purity zirconia powder and preparation method thereof

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115343625A (en) * 2022-10-17 2022-11-15 力高(山东)新能源技术股份有限公司 Power battery SOC estimation method based on error correction
CN116300410A (en) * 2023-05-25 2023-06-23 武汉理工大学 Corner optimization method and system for data-driven feedforward and feedback compensation
CN116300410B (en) * 2023-05-25 2023-08-22 武汉理工大学 Corner optimization method and system for data-driven feedforward and feedback compensation
CN117658208A (en) * 2023-12-20 2024-03-08 潮州市丰业新材料有限公司 High-purity zirconia powder and preparation method thereof
CN117658208B (en) * 2023-12-20 2024-06-04 潮州市丰业新材料有限公司 High-purity zirconia powder and preparation method thereof

Similar Documents

Publication Publication Date Title
CN110068774A (en) Estimation method, device and the storage medium of lithium battery health status
CN114779103A (en) Lithium ion battery SOC estimation method based on time-lag convolutional neural network
CN111007399B (en) Lithium battery state of charge prediction method based on improved generation countermeasure network
CN113030764B (en) Battery pack health state estimation method and system
CN110398697B (en) Lithium ion health state estimation method based on charging process
EP3786653B1 (en) Battery state estimation method
CN106026260A (en) SOC estimation method for series-wound battery pack having equalization circuit
CN104577242B (en) A kind of batteries management system and method
CN110146822A (en) A kind of Vehicular dynamic battery capacity On-line Estimation method based on constant-current charge process
CN112345939B (en) Lithium ion battery model parameter identification method based on continuous impulse response
CN109459699A (en) A kind of lithium-ion-power cell SOC method of real-time
CN111366848A (en) Battery health state prediction method based on PSO-ELM algorithm
CN112858916B (en) Battery pack state of charge estimation method based on model and data driving fusion
Zhang et al. Intelligent state of charge estimation of battery pack based on particle swarm optimization algorithm improved radical basis function neural network
Takyi-Aninakwa et al. An optimized long short-term memory-weighted fading extended Kalman filtering model with wide temperature adaptation for the state of charge estimation of lithium-ion batteries
CN112986830A (en) Lithium battery capacity estimation method based on convolution time memory neural network
CN110673037B (en) Battery SOC estimation method and system based on improved simulated annealing algorithm
WO2024000756A1 (en) Electrochemical model parameter identification method based on deep learning and heuristic algorithm
CN109613446A (en) A kind of aging of lithium battery detection method based on time series analysis
CN113655385A (en) Lithium battery SOC estimation method and device and computer readable storage medium
CN109655751A (en) A kind of method and system using Gaussian process regression estimates battery charging state
Takyi-Aninakwa et al. A hybrid probabilistic correction model for the state of charge estimation of lithium-ion batteries considering dynamic currents and temperatures
CN113820615A (en) Battery health degree detection method and device
CN113420444A (en) Lithium ion battery SOC estimation method based on parameter online identification
CN108446494B (en) Equalization algorithm for battery module or system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination