CN113805064A - Lithium ion battery pack health state prediction method based on deep learning - Google Patents

Lithium ion battery pack health state prediction method based on deep learning Download PDF

Info

Publication number
CN113805064A
CN113805064A CN202111098929.9A CN202111098929A CN113805064A CN 113805064 A CN113805064 A CN 113805064A CN 202111098929 A CN202111098929 A CN 202111098929A CN 113805064 A CN113805064 A CN 113805064A
Authority
CN
China
Prior art keywords
battery pack
prediction
lithium ion
ion battery
data
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
Application number
CN202111098929.9A
Other languages
Chinese (zh)
Other versions
CN113805064B (en
Inventor
王晓红
王立志
张钰
孙雅宁
林逸群
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.)
Beihang University
Original Assignee
Beihang 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 Beihang University filed Critical Beihang University
Priority to CN202111098929.9A priority Critical patent/CN113805064B/en
Publication of CN113805064A publication Critical patent/CN113805064A/en
Application granted granted Critical
Publication of CN113805064B publication Critical patent/CN113805064B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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]
    • G01R31/367—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]
    • G01R31/392—Determining battery ageing or deterioration, e.g. state of health
    • 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]
    • G01R31/396—Acquisition or processing of data for testing or for monitoring individual cells or groups of cells within a battery
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/01—Probabilistic graphical models, e.g. probabilistic networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Computational Mathematics (AREA)
  • Algebra (AREA)
  • Mathematical Analysis (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Secondary Cells (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)

Abstract

The invention discloses a lithium ion battery pack health state prediction method based on deep learning, which comprises the following steps of: s1, extracting the discharge characteristic parameters of the lithium ion battery pack, and carrying out correlation analysis on the discharge characteristic parameters; s2, obtaining an original feature data set according to the result of the correlation analysis, and performing feature dimension reduction; s3, constructing a node degradation data prediction model; s4, inputting the dimensionality reduction feature set into a node degradation data prediction model to obtain node degradation prediction data of the lithium ion battery pack; and S5, converting the node degradation prediction data into node state distribution prediction data, and acquiring the health state prediction result of the lithium ion battery pack based on the node state distribution prediction data. The method can realize accurate prediction of the health state of the lithium ion battery pack according to the processes of feature extraction, correlation analysis, feature dimension reduction, node state distribution prediction and system health state prediction.

Description

Lithium ion battery pack health state prediction method based on deep learning
Technical Field
The invention relates to the technical field of lithium ion battery state prediction, in particular to a lithium ion battery pack health state prediction method based on deep learning.
Background
The lithium ion battery is widely applied to energy power systems of equipment such as electric automobiles, hybrid ships, unmanned aerial vehicles and the like due to the characteristics of high specific energy, long cycle life, wide working temperature range and the like. Because the voltage of the single lithium ion battery is low, energy supply needs to be carried out in a battery pack mode in the power supply process, and therefore in actual use, hundreds of batteries form a complex energy supply system through various composite connection modes and various pack forming methods. The complete energy power system is formed by the single batteries, the combined connection battery pack and the complex battery pack, and is a typical multi-level system.
On the other hand, in the practical use process of the lithium ion battery pack, the consistency of the battery pack is difficult to maintain due to the strong dependence effect among batteries and the influence of the initial difference and the environmental stress of the batteries. Under the action of the dependency, the inconsistency will increasingly aggravate the unbalanced effect of the battery pack, so that the degradation process and the characteristic parameters between the batteries are mutually coupled and complicated, the mutual association relationship is dynamically changed, and the accurate prediction is difficult.
Disclosure of Invention
Aiming at the problems, the invention provides a lithium ion battery pack health state prediction method based on deep learning, which aims to solve the technical problems in the prior art, can completely perform corresponding analysis work according to the processes of test, feature extraction, correlation analysis, feature dimension reduction, node degradation data prediction, node state distribution prediction and system health state prediction, and realizes accurate prediction of the health state of the lithium ion battery pack.
In order to achieve the purpose, the invention provides the following scheme: the invention provides a lithium ion battery pack health state prediction method based on deep learning, which comprises the following steps of:
s1, extracting the discharge characteristic parameters of the lithium ion battery pack, and carrying out correlation analysis on the discharge characteristic parameters;
s2, obtaining an original feature data set according to the result of the correlation analysis, and performing feature dimension reduction on the original feature data set to obtain a dimension reduction feature set;
s3, obtaining the time sequence of the discharge condition of the lithium ion battery pack among the cycles according to the incidence relation of the discharge capacity among the cycles, and constructing a node degradation data prediction model based on the time sequence of the discharge condition of the lithium ion battery pack among the cycles;
s4, inputting the dimensionality reduction feature set into the node degradation data prediction model to obtain node degradation prediction data of the lithium ion battery pack;
and S5, converting the node degradation prediction data into node state distribution prediction data, and acquiring a health state prediction result of the lithium ion battery pack based on the node state distribution prediction data.
Preferably, the discharge characteristic parameters in S1 include discharge current, discharge voltage, discharge temperature, and discharge capacity data.
Preferably, the correlation analysis in S1 includes: and calculating the mean value, median and variance of the discharge characteristic parameters to complete statistical analysis, and performing related visual operation according to the result of the statistical analysis to complete correlation analysis.
Preferably, the acquiring process of the original feature data set in S2 is: and quantizing the result of the correlation analysis, obtaining a correlation coefficient matrix of the discharge characteristic parameters, removing the discharge characteristic parameters with the correlation coefficient less than 0.1 according to the correlation coefficient matrix, and obtaining an original characteristic data set.
Preferably, the feature dimension reduction process in S2 includes the following steps:
s2.1, carrying out gridding processing on the original characteristic data set to obtain a grid characteristic matrix;
s2.2, constructing a convolutional neural network model based on the grid characteristic matrix, and performing model optimization training on the convolutional neural network model;
and S2.3, obtaining dimension reduction characteristic parameters based on the convolutional neural network model after model optimization training, and finishing characteristic dimension reduction.
Preferably, the building of the node degradation data prediction model in S3 includes the following steps:
s3.1, acquiring a cycle degradation data set of the lithium ion battery pack, and acquiring a model data set based on the cycle degradation data set;
s3.2, constructing a node degradation data prediction model;
and S3.3, performing parameter optimization training on the node degradation data prediction model based on the model data set to complete construction of the node degradation data prediction model.
Preferably, the process of obtaining the health status prediction result of the lithium ion battery pack in S5 includes the following steps:
s5.1, discretizing the node degradation prediction data to finish multi-state division of the degradation process of the lithium ion battery pack;
s5.2, constructing a Bayesian neural network mixed model;
and S5.3, acquiring node state distribution prediction data based on the Bayesian neural network mixed model, and acquiring a health state prediction result of the lithium ion battery pack according to the node state distribution prediction data.
The invention discloses the following technical effects:
the method utilizes a multi-level system prediction modeling method, establishes a battery pack system Bayesian network model through multi-characteristic parameters of battery charging and discharging, and completely performs corresponding analysis work according to the flows of characteristic extraction-correlation analysis-characteristic dimension reduction-node state distribution prediction-system health state prediction, thereby realizing accurate prediction of the health state of the lithium ion battery pack.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic view of a battery pack testing platform according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a parallel-series battery pack according to an embodiment of the present invention;
FIG. 4 is a schematic cross-sectional view of a cycle according to an embodiment of the present invention; wherein (a) is the discharge cycle 5; (b) cycle 295 for discharge;
FIG. 5 is a schematic diagram of feature dimension reduction according to an embodiment of the present invention;
FIG. 6 is an integrated node state probability distribution prediction curve according to an embodiment of the present invention;
fig. 7 is a bayesian network model of a battery system according to an embodiment of the invention;
fig. 8 shows the prediction result of the system state of the lithium ion battery pack according to the embodiment of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Referring to fig. 1, the present embodiment provides a method for predicting the health status of a lithium ion battery pack based on deep learning, which includes the following steps:
and S1, extracting the discharge characteristic parameters of the lithium ion battery pack, and carrying out correlation analysis on the discharge characteristic parameters.
The discharge characteristic parameters comprise discharge current, discharge voltage, discharge temperature and discharge capacity data. The correlation analysis process is as follows: and calculating the mean value, median and variance of the discharge characteristic parameters to complete statistical analysis, and performing related visual operation according to the result of the statistical analysis to complete correlation analysis.
And S2, acquiring an original characteristic data set according to the result of the correlation analysis, and performing characteristic dimension reduction on the original characteristic data set through a convolutional neural network to obtain a dimension reduction characteristic set.
The acquisition process of the original characteristic data set comprises the following steps: and quantifying the result of the correlation analysis, obtaining a correlation coefficient matrix of the discharge characteristic parameters, removing the discharge characteristic parameters with the correlation coefficient less than 0.1 according to the correlation coefficient matrix, and obtaining an original characteristic data set.
The characteristic dimension reduction process comprises the following steps:
s2.1, carrying out gridding processing on the original characteristic data set to obtain a grid characteristic matrix;
s2.2, constructing a convolutional neural network model based on the grid characteristic matrix, and performing model optimization training on the convolutional neural network model;
and S2.3, obtaining dimension reduction characteristic parameters based on the convolutional neural network model after model optimization training, and finishing characteristic dimension reduction.
S3, obtaining the time sequence of the discharge condition of the lithium ion battery pack among the cycles according to the incidence relation of the discharge capacity among the cycles, and constructing a node degradation data prediction model based on the time sequence of the discharge condition of the lithium ion battery pack among the cycles.
The construction of the node degradation data prediction model comprises the following steps:
s3.1, acquiring a cycle degradation data set of the lithium ion battery pack, and acquiring a model data set based on the cycle degradation data set;
s3.2, constructing a node degradation data prediction model;
and S3.3, performing parameter optimization training on the node degradation data prediction model based on the model data set to complete construction of the node degradation data prediction model.
And S4, inputting the dimension reduction feature set into the node degradation data prediction model to obtain node degradation prediction data of the lithium ion battery pack.
And S5, converting the node degradation prediction data into node state distribution prediction data, and acquiring the health state prediction result of the lithium ion battery pack based on the node state distribution prediction data.
S5.1, discretizing the node degradation prediction data to complete multi-state division of the degradation process of the lithium ion battery pack;
s5.2, constructing a Bayesian neural network mixed model based on a deep learning and program implementation framework;
and S5.3, acquiring node state distribution prediction data based on the Bayesian neural network mixed model, and acquiring a health state prediction result of the lithium ion battery pack according to the node state distribution prediction data.
In the cycle degradation test of the lithium ion battery pack in this embodiment, a commercial ICR18650 lithium ion battery is adopted as a tested battery, four batteries are connected in parallel and then in series to form the battery pack, and a cycle degradation test platform and a test system of the battery pack are schematically shown in fig. 2. The structure of the parallel-series battery is schematically shown in fig. 3, wherein Cell _1, Cell _2, Cell _3 and Cell _4 are the numbers of the single batteries.
In order to evaluate and predict the health state of the battery pack, the degradation test is mainly used for evaluating the performance of each discharge characteristic parameter of the battery pack, corresponding measurement parameters comprise characteristic parameters of discharge current, discharge voltage, discharge temperature and the like of the battery pack and internal single batteries, and the overall cyclic discharge capacity of the battery pack is measured as the health state index of the battery pack. For the conversion content part of the degradation data to the state distribution, the target is a parallel module integration node (hereinafter referred to as a parallel module) composed of Cell _1 and Cell _ 2.
The parallel module cycle degradation test data set comprises 320 cycles of discharge data, the characteristic parameters comprise charge and discharge current, charge and discharge voltage and charge and discharge temperature of each cycle of the single batteries in the parallel module and the parallel module, and the discharge capacity data of the whole parallel module, and the cycle section schematic diagram of the parallel module is shown in fig. 4.
In the experiment, the dependence of the parallel modules in the discharge stage is particularly obvious, the discharge characteristic difference of the battery pack is obvious along with the cyclic degradation, the unbalanced influence is obviously strengthened, the discharge characteristic can reflect the current discharge performance of the battery pack, the parallel modules can be well used as the input parameter of the health state of the battery pack, and the discharge capacity is the main index reflecting the health state of the battery pack. On the basis, the discharge stage of the circulation profile of the parallel module is intercepted, and the characteristics of the discharge stage and the relevant properties of the change of the overall health state of the system are analyzed.
The characteristic parameters also change obviously with the progress of the degradation state: along with the progress of the cyclic degradation process, the fluctuation of current parameters is obviously increased, the inconsistency among the batteries is also obviously strengthened, and the influence of the inconsistency is also spread to the whole discharge process from the discharge end stage; for voltage parameters, due to the characteristics of a parallel structure, the change of the early and later degradation stages is not obvious, and the correlation between the voltage parameters and the state of a parallel module needs to be further analyzed along with the enhancement of the degradation fluctuation of the battery; the temperature parameters are also obviously changed, firstly, the temperature mean value is obviously improved, secondly, the inconsistency is obviously enhanced, and the difference of the two batteries is increased. On the other hand, as can also be seen from the cross-sectional views of the discharge characteristics, the change of the battery health state also has certain position characteristics in the change of the characteristic parameters, for example, from the view of current and temperature data, the degradation process of the battery mainly diffuses from the end stage of discharge to the front, the capture of such characteristic information is realized, the effect of reducing the dimension of the characteristic is improved, and the effect is also one of the main effects of the convolution kernel in the convolutional neural network.
And the statistical characteristics of the discharge parameters can also effectively reflect the health state of the battery pack, for example, the mean value of the temperature can reflect the change trend of the internal resistance of the battery, the variance of the current and the voltage can reflect the discharge stability of the battery, the measured discharge characteristic parameters of the plurality of parallel modules are subjected to statistical analysis, the mean value, the median and the variance of each parameter are calculated, and relevant visual operation is carried out.
Each statistical parameter and the degradation process of the parallel module have strong correlation property, in order to quantify the correlation relationship, a characteristic parameter correlation coefficient matrix is calculated, the characteristics with the correlation number smaller than 0.1 are removed according to the correlation matrix, and the remaining characteristics and the previous discharge current, discharge voltage and discharge temperature are used as the original characteristic set to be input, as shown in tables 1 and 2.
TABLE 1
Discharge characteristics Characteristic data type
Cell _1 discharge current Zhang Liang
Cell _1 discharge voltage Zhang Liang
Cell _1 discharge temperature Zhang Liang
Cell _2 discharge current Zhang Liang
Cell _2 discharge voltage Zhang Liang
Cell _2 discharge temperature Zhang Liang
Cell _1 median discharge current Scalar quantity
Cell _1 discharge current mean value Scalar quantity
Cell _1 discharge current variance Scalar quantity
Cell _2 median of discharge current Scalar quantity
TABLE 2
Discharge characteristics Characteristic data type
Cell _2 mean discharge current Scalar quantity
Cell _2 discharge current variance Scalar quantity
Cell _1 discharge voltage median Scalar quantity
Cell _1 discharge voltage mean value Scalar quantity
Cell _2 median discharge voltage Scalar quantity
Cell _2 discharge voltage mean value Scalar quantity
Cell _1 discharge temperature variance Scalar quantity
Cell _1 discharge temperature mean value Scalar quantity
Cell _2 discharge temperature variance Scalar quantity
Cell _2 mean discharge temperature Scalar quantity
Further analyzing the correlation coefficient matrix, it can be seen that the correlation of the characteristic parameters to the health state of the parallel module is basically high, which means that the extracted battery charge-discharge characteristics can be used for predicting the health state of the battery pack, but at the same time, parameters with low correlation (less than 0.1) such as temperature median, voltage variance, and the like exist, and strong correlation properties exist among the multi-characteristic parameters, so that if the model is directly input, weight deviation is caused, the model prediction accuracy is reduced, and the overlapping problem of multi-characteristic data is brought, so that further feature dimension reduction work is required.
In order to obtain higher-quality input features to improve the model prediction accuracy, for the obtained original feature data set, feature dimension reduction is performed on the discharge features of the lithium ion battery pack by using a convolutional neural network based on a feature dimension reduction method, as shown in fig. 5.
In order to analyze and evaluate the result of the dimension reduction of the model characteristics, the health state of the parallel modules is subjected to multi-state division, the health, the degradation and the failure are used as prediction labels, and the Adam algorithm is used as an optimization algorithm to perform optimization training of model parameters.
For the evaluation of the multi-feature dimension reduction effect, the embodiment evaluates and analyzes the feature dimension reduction result from two aspects of dimension reduction feature visualization and model classification performance. And carrying out visualization processing on the first three dimensionality reduction main components of the dimensionality reduction result. Original multi-feature data are subjected to nonlinear combination between features through a convolutional neural network to obtain dimension reduction feature parameters. Analogizing to principal component analysis, regarding a visualization result, three coordinate axes in the image represent the first three principal component dimensions of the dimension reduction feature result, each image represents a cycle process of the battery pack, and the change trend of the curve represents the change condition of the dimension reduction feature of the parallel module in the current discharge cycle process.
The model prediction accuracy is used as an evaluation standard, and a model prediction result acc is shown as follows:
Figure BDA0003270125340000101
wherein m represents the number of samples for which a prediction is correct; n represents the total sample size. The model prediction results are shown in table 3.
TABLE 3
Total number of samples (n) Predicting correct sample number (m) Rate of accuracy
320 309 96.6％
The prediction precision result of the model is 96.6%, the model has good prediction performance, and the dimensionality reduction characteristic obtained by the convolutional neural network can well represent the health state of the system. And expressing the degradation process of each cycle by using three main components obtained by dimensionality reduction extraction. With the progress of the cyclic degradation process, the disorder degree of the discharge process of the parallel modules is obviously increased, and from the aspect of pattern recognition, the health state change of the battery pack is effectively recognized through the characteristics obtained through the dimension reduction of the convolutional neural network.
On the basis, the degradation characteristic parameters of the parallel modules have strong correlation with the change of the health state of the battery pack, and the degradation state of the battery pack can be well represented. The cycle discharge capacity is used as a main index for representing the health state of the battery pack, so the modeling and prediction work is mainly carried out on the cycle discharge capacity of the battery pack.
Considering that the health state of the battery pack has continuity in time sequence dimension, and the discharge capacity among cycles has an incidence relation, namely the time sequence of the discharge condition among the cycles of the battery pack, a mixed model of a convolutional neural network and an LSTM cyclic neural network is established, and the cyclic discharge capacity of the parallel modules is predicted.
And predicting the discharge capacity of the parallel modules by using the CNN + LSTM mixed model, wherein the model has a good prediction effect on the discharge capacity of the battery pack. In terms of Root Mean Square Error (RMSE), absolute error (MAE) and R2The values were used as model evaluation indices, and the corresponding model prediction set evaluation results are shown in table 4.
TABLE 4
Evaluation index RMSE(mAh) MAE(mAh) R2-value
Predicted results 9.90 8.68 0.87
The results in Table 4 show that the prediction error of the model to the capacity can reach about 10mAh, R2Value of>0.85, has good prediction performance.
Wherein, RMSE, MAE, and R2The calculation formulas of the values are respectively as follows:
Figure BDA0003270125340000111
Figure BDA0003270125340000121
Figure BDA0003270125340000122
in the formula, yiRepresenting the ith true value;
Figure BDA0003270125340000123
a prediction value representing the i-th prediction; i represents the number of measurements, i is 1,2, … …, m.
On the basis, in order to clearly evaluate the effectiveness of the deep learning model prediction, the present embodiment continues to further evaluate the prediction accuracy of the proposed model. The stability of model prediction is analyzed, and the case diagram is adopted to carry out prediction accuracy dispersion statistics in the embodiment. Firstly, converting an original prediction result into a relative error calculation, wherein a calculation formula of the relative error is as follows:
Figure BDA0003270125340000124
in the formula (I), the compound is shown in the specification,
Figure BDA0003270125340000125
representing the true value; y represents a predicted value.
Through the box line analysis result, the prediction result of the deep learning prediction model on the health state degradation quantity of the parallel modules is low in volatility, few in outlier and good in performance.
The prediction of the battery health state degradation process is realized, but to realize the inference of the lithium ion battery pack system state, the degradation data of the battery pack needs to be converted into state distribution data, and the degradation data can be converted into node state distribution data by using a Bayesian neural network.
The degradation process of the battery pack is first state-divided. Continuous data acquired by the base layer component at different moments are discretized by fuzzy numbers based on a fuzzy theory, namely multi-state division of data in the degradation process. The more finely the state division of the capacity degradation data of the lithium ion battery is, the more accurately the state of the capacity degradation data of the lithium ion battery can be grasped, however, as the number of the states increases, the complexity of calculation increases in an exponential level, and for a lithium ion battery pack system, trapezoidal fuzzy numbers are suitable for dividing the continuous data of the capacity degradation data into 3 states (healthy, degraded, invalid/normal, failed) or 4 states (healthy, slightly degraded, severely degraded, invalid/normal, slope degraded, serious degraded, failed) for research.
And discretizing the normalized capacity data according to a multi-state division method, comprehensively calculating complexity and accuracy according to the network node state established in the battery degradation process, and dividing the node state of the established battery pack system network into 3 states (health, degradation, failure/normal, degraded and failure) to obtain the multi-state division result of the battery pack system and the single node.
By utilizing the characteristics of a deep learning and program implementation framework, on the basis of the established CNN + LSTM mixed model, a Bayesian neural network layer is defined, the prior distribution of hidden neurons is set to be Gaussian distribution, the full-connection network layer of the original model is replaced, the output dimension of the network model is set according to the health state division condition, and the CNN + LSTM + BNN (Bayesian neural network) mixed model is established.
On the basis of state division, a Bayesian neural network model is established, and a state probability distribution prediction curve of an integration node is shown in FIGS. 6-8. As can be seen from fig. 8, the node starts to transition from the healthy state to the degraded state around 100 cycles of the integration node, and the probability of the failure state suddenly increases around 240 cycles, so that the battery parallel module rapidly enters the degraded state.
The invention discloses the following technical effects:
the method utilizes a multi-level system prediction modeling method, establishes a battery pack system Bayesian network model through multi-characteristic parameters of battery charging and discharging, and completely performs corresponding analysis work according to the flows of characteristic extraction-correlation analysis-characteristic dimension reduction-node state distribution prediction-system health state prediction, thereby realizing accurate prediction of the health state of the lithium ion battery pack.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the present invention in its spirit and scope. Are intended to be covered by the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (7)

1. The lithium ion battery pack health state prediction method based on deep learning is characterized by comprising the following steps of:
s1, extracting the discharge characteristic parameters of the lithium ion battery pack, and carrying out correlation analysis on the discharge characteristic parameters;
s2, obtaining an original feature data set according to the result of the correlation analysis, and performing feature dimension reduction on the original feature data set to obtain a dimension reduction feature set;
s3, obtaining the time sequence of the discharge condition of the lithium ion battery pack among the cycles according to the incidence relation of the discharge capacity among the cycles, and constructing a node degradation data prediction model based on the time sequence of the discharge condition of the lithium ion battery pack among the cycles;
s4, inputting the dimensionality reduction feature set into the node degradation data prediction model to obtain node degradation prediction data of the lithium ion battery pack;
and S5, converting the node degradation prediction data into node state distribution prediction data, and acquiring a health state prediction result of the lithium ion battery pack based on the node state distribution prediction data.
2. The deep learning-based lithium ion battery pack state of health prediction method of claim 1, wherein the discharge characteristic parameters in S1 include discharge current, discharge voltage, discharge temperature, discharge capacity data.
3. The deep learning based lithium ion battery pack state of health prediction method of claim 1, wherein the correlation analysis in S1 is performed by: and calculating the mean value, median and variance of the discharge characteristic parameters to complete statistical analysis, and performing related visual operation according to the result of the statistical analysis to complete correlation analysis.
4. The method for predicting the health status of a lithium ion battery pack based on deep learning of claim 1, wherein the step of acquiring the original characteristic data set in S2 is as follows: and quantizing the result of the correlation analysis, obtaining a correlation coefficient matrix of the discharge characteristic parameters, removing the discharge characteristic parameters with the correlation coefficient less than 0.1 according to the correlation coefficient matrix, and obtaining an original characteristic data set.
5. The method for predicting the state of health of a lithium ion battery pack based on deep learning of claim 1, wherein the feature dimension reduction process in the step S2 comprises the following steps:
s2.1, carrying out gridding processing on the original characteristic data set to obtain a grid characteristic matrix;
s2.2, constructing a convolutional neural network model based on the grid characteristic matrix, and performing model optimization training on the convolutional neural network model;
and S2.3, obtaining dimension reduction characteristic parameters based on the convolutional neural network model after model optimization training, and finishing characteristic dimension reduction.
6. The deep learning-based lithium ion battery pack state of health prediction method of claim 1, wherein the construction of the node degradation data prediction model in S3 comprises the following steps:
s3.1, acquiring a cycle degradation data set of the lithium ion battery pack, and acquiring a model data set based on the cycle degradation data set;
s3.2, constructing a node degradation data prediction model;
and S3.3, performing parameter optimization training on the node degradation data prediction model based on the model data set to complete construction of the node degradation data prediction model.
7. The method for predicting the health status of the lithium ion battery pack based on deep learning of claim 1, wherein the step of obtaining the health status prediction result of the lithium ion battery pack in S5 comprises the following steps:
s5.1, discretizing the node degradation prediction data to finish multi-state division of the degradation process of the lithium ion battery pack;
s5.2, constructing a Bayesian neural network mixed model;
and S5.3, acquiring node state distribution prediction data based on the Bayesian neural network mixed model, and acquiring a health state prediction result of the lithium ion battery pack according to the node state distribution prediction data.
CN202111098929.9A 2021-09-18 2021-09-18 Lithium ion battery pack health state prediction method based on deep learning Active CN113805064B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111098929.9A CN113805064B (en) 2021-09-18 2021-09-18 Lithium ion battery pack health state prediction method based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111098929.9A CN113805064B (en) 2021-09-18 2021-09-18 Lithium ion battery pack health state prediction method based on deep learning

Publications (2)

Publication Number Publication Date
CN113805064A true CN113805064A (en) 2021-12-17
CN113805064B CN113805064B (en) 2022-09-20

Family

ID=78939650

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111098929.9A Active CN113805064B (en) 2021-09-18 2021-09-18 Lithium ion battery pack health state prediction method based on deep learning

Country Status (1)

Country Link
CN (1) CN113805064B (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114280491A (en) * 2021-12-23 2022-04-05 中山大学 Retired battery residual capacity estimation method based on active learning
CN114325449A (en) * 2022-01-05 2022-04-12 北京理工大学深圳汽车研究院(电动车辆国家工程实验室深圳研究院) Lithium ion battery SOH prediction method
CN114462259A (en) * 2022-04-13 2022-05-10 长安新能源南京研究院有限公司 SOC-based battery remaining life prediction method and system, automobile and medium
CN114636933A (en) * 2022-05-23 2022-06-17 长沙矿冶研究院有限责任公司 Composite neural network-based retired power battery capacity detection method and system
CN114781614A (en) * 2022-04-27 2022-07-22 电子科技大学 Lithium ion battery health state prediction method
CN115508711A (en) * 2022-10-10 2022-12-23 南京工程学院 Electric vehicle power lithium battery residual life prediction method based on deep network model
CN115598557A (en) * 2022-08-26 2023-01-13 广东工业大学（Cn） Lithium battery SOH estimation method based on constant voltage charging current
CN116388353A (en) * 2023-06-06 2023-07-04 宁波齐云新材料技术有限公司 Power supply system for lithium battery pack and control method
CN117148165A (en) * 2023-09-15 2023-12-01 东莞市言科新能源有限公司 Testing and analyzing method and system for polymer lithium ion battery

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106526486A (en) * 2016-08-30 2017-03-22 郑州轻工业学院 Construction method for lithium battery health life model
CN106772064A (en) * 2016-11-25 2017-05-31 广东电网有限责任公司电力科学研究院 A kind of health state of lithium ion battery Forecasting Methodology and device
CN108287317A (en) * 2018-01-24 2018-07-17 上海电气分布式能源科技有限公司 Power of battery prediction model generation method and system, power forecasting method and system
US20180306868A1 (en) * 2017-04-25 2018-10-25 Korea University Of Technology And Education Industry-University Cooperation Method for predicting state of health of battery based on numerical simulation data
EP3736587A1 (en) * 2019-05-08 2020-11-11 Tata Consultancy Services Limited A method and a system for estimation of remaining useful life in lithium based batteries
CN112630662A (en) * 2020-12-30 2021-04-09 南京航空航天大学 Power battery SOH estimation method based on data driving and multi-parameter fusion
CN113391209A (en) * 2021-05-26 2021-09-14 江苏小牛电动科技有限公司 Method, device and system for predicting health state of battery and battery

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106526486A (en) * 2016-08-30 2017-03-22 郑州轻工业学院 Construction method for lithium battery health life model
CN106772064A (en) * 2016-11-25 2017-05-31 广东电网有限责任公司电力科学研究院 A kind of health state of lithium ion battery Forecasting Methodology and device
US20180306868A1 (en) * 2017-04-25 2018-10-25 Korea University Of Technology And Education Industry-University Cooperation Method for predicting state of health of battery based on numerical simulation data
CN108287317A (en) * 2018-01-24 2018-07-17 上海电气分布式能源科技有限公司 Power of battery prediction model generation method and system, power forecasting method and system
EP3736587A1 (en) * 2019-05-08 2020-11-11 Tata Consultancy Services Limited A method and a system for estimation of remaining useful life in lithium based batteries
CN112630662A (en) * 2020-12-30 2021-04-09 南京航空航天大学 Power battery SOH estimation method based on data driving and multi-parameter fusion
CN113391209A (en) * 2021-05-26 2021-09-14 江苏小牛电动科技有限公司 Method, device and system for predicting health state of battery and battery

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114280491B (en) * 2021-12-23 2024-01-05 中山大学 Retired battery residual capacity estimation method based on active learning
CN114280491A (en) * 2021-12-23 2022-04-05 中山大学 Retired battery residual capacity estimation method based on active learning
CN114325449B (en) * 2022-01-05 2023-09-19 北京理工大学深圳汽车研究院(电动车辆国家工程实验室深圳研究院) SOH prediction method for lithium ion battery
CN114325449A (en) * 2022-01-05 2022-04-12 北京理工大学深圳汽车研究院(电动车辆国家工程实验室深圳研究院) Lithium ion battery SOH prediction method
CN114462259A (en) * 2022-04-13 2022-05-10 长安新能源南京研究院有限公司 SOC-based battery remaining life prediction method and system, automobile and medium
CN114781614A (en) * 2022-04-27 2022-07-22 电子科技大学 Lithium ion battery health state prediction method
CN114781614B (en) * 2022-04-27 2023-09-15 电子科技大学 Lithium ion battery health state prediction method
CN114636933B (en) * 2022-05-23 2022-09-13 长沙矿冶研究院有限责任公司 Composite neural network-based retired power battery capacity detection method and system
CN114636933A (en) * 2022-05-23 2022-06-17 长沙矿冶研究院有限责任公司 Composite neural network-based retired power battery capacity detection method and system
CN115598557A (en) * 2022-08-26 2023-01-13 广东工业大学（Cn） Lithium battery SOH estimation method based on constant voltage charging current
CN115598557B (en) * 2022-08-26 2023-08-25 广东工业大学 Lithium battery SOH estimation method based on constant-voltage charging current
CN115508711A (en) * 2022-10-10 2022-12-23 南京工程学院 Electric vehicle power lithium battery residual life prediction method based on deep network model
CN116388353A (en) * 2023-06-06 2023-07-04 宁波齐云新材料技术有限公司 Power supply system for lithium battery pack and control method
CN116388353B (en) * 2023-06-06 2023-09-08 宁波齐云新材料技术有限公司 Power supply system for lithium battery pack and control method
CN117148165A (en) * 2023-09-15 2023-12-01 东莞市言科新能源有限公司 Testing and analyzing method and system for polymer lithium ion battery
CN117148165B (en) * 2023-09-15 2024-04-12 东莞市言科新能源有限公司 Testing and analyzing method and system for polymer lithium ion battery

Also Published As

Publication number Publication date
CN113805064B (en) 2022-09-20

Similar Documents

Publication Publication Date Title
CN113805064B (en) Lithium ion battery pack health state prediction method based on deep learning
CN111443294B (en) Method and device for indirectly predicting remaining life of lithium ion battery
Li et al. Lithium-ion battery state of health monitoring based on ensemble learning
CN113777496B (en) Lithium ion battery residual life prediction method based on time convolution neural network
CN110221225B (en) Spacecraft lithium ion battery cycle life prediction method
CN108535656A (en) Lithium ion battery remaining life prediction technique and system based on PCA-NARX neural networks
Zhang et al. Intelligent state of charge estimation of battery pack based on particle swarm optimization algorithm improved radical basis function neural network
CN107576920A (en) A kind of satellite power supply health degree computational methods based on fuzzy multi-attribute evaluation
EP4113139A1 (en) Lithium battery soc estimation method and apparatus, and computer-readable storage medium
CN112834927A (en) Lithium battery residual life prediction method, system, device and medium
Wang et al. Health diagnosis for lithium-ion battery by combining partial incremental capacity and deep belief network during insufficient discharge profile
Zheng et al. State of health estimation for lithium battery random charging process based on CNN-GRU method
CN115201686B (en) Lithium ion battery health state assessment method under incomplete charge and discharge data
CN112904219A (en) Big data-based power battery health state prediction method
CN112115648A (en) Transformer top layer oil temperature prediction method based on improved deep learning method
CN115598557B (en) Lithium battery SOH estimation method based on constant-voltage charging current
CN115409263A (en) Method for predicting remaining life of lithium battery based on gating and attention mechanism
CN114280490B (en) Lithium ion battery state of charge estimation method and system
Cao et al. A flexible battery capacity estimation method based on partial voltage curves and polynomial fitting
Pau et al. Microcontroller architectures for battery state of charge prediction with tiny neural networks
CN116840724A (en) Lithium ion battery health condition estimation and residual service life prediction method
CN116908727A (en) Multi-feature combination-based energy storage power station lithium battery capacity estimation method
CN116736172A (en) Lithium battery pack health state evaluation method based on generation type countermeasure network
CN113255215B (en) Lithium battery health state estimation method based on voltage segments
CN114355218A (en) Lithium ion battery charge state prediction method based on multi-feature quantity screening

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