CN110222431B - Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion - Google Patents

Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion Download PDF

Info

Publication number
CN110222431B
CN110222431B CN201910502496.5A CN201910502496A CN110222431B CN 110222431 B CN110222431 B CN 110222431B CN 201910502496 A CN201910502496 A CN 201910502496A CN 110222431 B CN110222431 B CN 110222431B
Authority
CN
China
Prior art keywords
lithium ion
ion battery
capacity
model
battery
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910502496.5A
Other languages
Chinese (zh)
Other versions
CN110222431A (en
Inventor
刘大同
彭喜元
李律
宋宇晨
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.)
Harbin Institute of Technology
Original Assignee
Harbin Institute of Technology
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 Harbin Institute of Technology filed Critical Harbin Institute of Technology
Priority to CN201910502496.5A priority Critical patent/CN110222431B/en
Publication of CN110222431A publication Critical patent/CN110222431A/en
Application granted granted Critical
Publication of CN110222431B publication Critical patent/CN110222431B/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
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Secondary Cells (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)
  • Tests Of Electric Status Of Batteries (AREA)

Abstract

A lithium ion battery residual life prediction method based on the fusion of a gated cyclic unit neural network and a Kalman filtering model relates to the technical field of lithium ion battery health state detection. The invention aims to solve the problems of poor fitting capability and low adaptability to different working states in a nonlinear degradation process in the existing lithium ion battery residual life prediction method based on a fusion model. According to the invention, through establishing a GRU-RNN deep network model and utilizing the strong characteristic extraction capability of the GRU deep learning model on a time sequence, the capacity degradation characteristics of the lithium ion battery are extracted, so that a more accurate battery capacity prediction model is obtained, and finally, noise is reduced through a KF filtering method, and a more accurate prediction value is obtained.

Description

Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion
Technical Field
The invention belongs to the technical field of lithium ion battery health state detection, and particularly relates to a battery life prediction technology.
Background
The current methods for predicting the Remaining Life of lithium ion batteries (RUL) are roughly classified into two types, namely physical model-based methods and data-driven model-based methods. The data driving method does not need to clearly determine the degradation mechanism of the battery, so that the related research is more intensive. The data driving method comprises a statistical data driving method based on statistical filtering and a data driving method based on a machine learning method. The lithium ion battery degradation model used by the data driving method based on the statistical filtering is single, and the adaptability to the service life prediction problem of different types of batteries and different use conditions is poor. The data driving method based on the machine learning method only focuses on the correlation among data, and the characteristics of the lithium battery to be tested are less considered. At present, battery residual life prediction methods which integrate a statistical filtering method and a data-driven model are widely researched, but the current methods have the problems of poor fitting capability in a nonlinear degradation process, low adaptability to different working states and the like.
Disclosure of Invention
The invention provides a lithium ion battery residual life prediction method based on gate controlled cycle Unit (GRU) neural network and Kalman Filter (KF) model fusion, aiming at solving the problems of poor fitting capability and low adaptability to different working states in the existing lithium ion battery residual life prediction method based on a fusion model. In a state space, the fusion of the two heterogeneous models is realized, and the accuracy of a prediction result is improved.
The lithium ion battery residual life prediction method based on the fusion of the gated cyclic unit neural network and the Kalman filtering model comprises the following steps:
step one, constructing a data set by using battery capacity data of a lithium ion battery for training in each charge-discharge cycle, and taking the data in the data set as training data;
step two, constructing a training set of the GRU model by using the training data;
substituting the training set into the GRU model, and training the network parameters in the GRU model by using a BP algorithm to obtain a trained GRU model;
constructing an input vector by using battery capacity data in the kth charging and discharging period of the lithium ion battery to be predicted;
substituting the input vector into the GRU model obtained in the step three to obtain a predicted value of the battery capacity of the lithium ion battery in the (k + 1) th charging and discharging period to be predicted;
step six, judging whether the predicted value obtained in the step five is less than or equal to 80% of the rated capacity of the lithium ion battery to be predicted or not, if yes, executing the step seven, otherwise, enabling k to be k +1, and then returning to the step four;
and step seven, taking the difference value between the charge-discharge period corresponding to the predicted value and the actual charge-discharge period of the battery as the residual life of the lithium ion battery to be predicted.
The training set of the GRU model in the second step is represented as [ xtrain, ytrain ], where:
Figure BDA0002090699210000021
in the above formula, CaplThe capacity of the first battery is shown,
Figure BDA0002090699210000022
represents the number of charge/discharge cycles corresponding to the deterioration of the capacity of the first battery to 80% of the rated capacity, wherein l and n are positive integers
Figure BDA0002090699210000023
GRU model includes an update gate ztAnd a reset gate rtLet the t-th column data in xtrain be xtThen, the internal propagation formula of the GRU model is:
zt＝σ(Wz·[ht-1,xt]) (1)
rt＝σ(Wr·[ht-1,xt]) (2)
Figure BDA0002090699210000024
Figure BDA0002090699210000025
yt＝σ(Wo·ht) (5)
wherein the operation symbol portion is]Representing the connection of the two vector endings, the product of the matrix,
Figure BDA0002090699210000026
Of the parameter part
Figure BDA0002090699210000027
Network parameters representing the GRU model,
Figure BDA0002090699210000028
And htAll represent intermediate variables, ytAn estimate of the GRU model is indicated.
The input vector described in step four is expressed as follows:
{Captest(k-n+1),Captest(k-n+2),...,Captest(k)} (6)
Captest(k) and the observed value represents the real capacity of the lithium ion battery in the k-th cycle.
Aiming at the problem that the long-term prediction error is large in the prediction of the remaining life of the battery, the method for predicting the remaining life of the lithium ion battery based on the fusion of the gated cyclic unit neural network and the Kalman filtering model can accurately predict the remaining life of different stages. According to the method, the capacity degradation characteristics of the lithium ion battery are extracted by establishing a GRU-RNN deep network model and utilizing the strong characteristic extraction capability of the GRU deep learning model on a time sequence, so that a more accurate battery capacity prediction model is obtained, finally, noise is reduced by a KF filtering method, and a more accurate prediction value is obtained. The maximum error of the residual life prediction result of the method is 11 periods, the average absolute value error is less than 0.06, the root mean square error is less than 0.016, and experimental results prove that the method can effectively predict the residual life of the lithium ion battery.
Drawings
Fig. 1 is a graph of the predicted results.
Detailed Description
The first embodiment is as follows: referring to fig. 1, specifically describing the present embodiment, in the lithium ion battery remaining life prediction method based on the gated cycle unit neural network and kalman filter model fusion described in the present embodiment, a 18650 battery with a rated capacity of 2200mAh is used for testing, and the test specifically includes the following steps:
step one, constructing a data set by using battery capacity data of the lithium ion battery for training in each charge-discharge cycle, and taking the data in the data set as training data.
Step two, constructing a training set of the GRU model by using the training data, wherein the training set is represented as [ xtrain, ytrain ], and comprises the following steps:
Figure BDA0002090699210000041
in the above formula, CaplThe capacity of the first battery is shown,
Figure BDA0002090699210000042
represents the number of charge/discharge cycles corresponding to the deterioration of the capacity of the first battery to 80% of the rated capacity, wherein l and n are positive integers
Figure BDA0002090699210000043
Substituting the training set into the GRU model, and training network parameters in the GRU model by using a BP (Error Back Propagation) algorithm to obtain a trained GRU model; GRU model includes an update gate ztAnd a reset gate rtLet the t-th column data in xtrain be xtThen, the internal propagation formula of the GRU model is:
zt＝σ(Wz·[ht-1,xt]) (1)
rt＝σ(Wr·[ht-1,xt]) (2)
Figure BDA0002090699210000044
Figure BDA0002090699210000045
yt＝σ(Wo·ht) (5)
wherein the operation symbol portion is]Representing the connection of the two vector endings, the product of the matrix,
Figure BDA0002090699210000046
Of the parameter part
Figure BDA0002090699210000047
Network parameters representing the GRU model,
Figure BDA0002090699210000048
And htAll represent intermediate variables, ytAn estimate of the GRU model is indicated.
Step four, constructing an input vector by using the battery capacity data in the kth charging and discharging period of the lithium ion battery to be predicted, wherein the input vector is expressed as follows:
{Captest(k-n+1),Captest(k-n+2),...,Captest(k)} (6)
Captest(k) and the observed value represents the real capacity of the lithium ion battery in the k-th cycle.
And step five, substituting the input vector into the GRU model obtained in the step three to obtain a predicted value of the battery capacity of the lithium ion battery in the (k + 1) th charging and discharging period to be predicted.
And step six, judging whether the predicted value obtained in the step five is less than or equal to 80% of the rated capacity of the lithium ion battery to be predicted, if so, executing the step seven, otherwise, enabling k to be k +1, and then returning to the step four.
And step seven, taking the difference value between the charge-discharge period corresponding to the predicted value and the actual charge-discharge period of the battery as the residual life of the lithium ion battery to be predicted.

Claims (1)

1. The lithium ion battery residual life prediction method based on the integration of the gated cyclic unit neural network and the Kalman filtering model is characterized in that,
step one, constructing a data set by using battery capacity data of a lithium ion battery for training in each charge-discharge cycle, and taking the data in the data set as training data;
step two, constructing a training set of the GRU model by using the training data;
substituting the training set into the GRU model, and training the network parameters in the GRU model by using a BP algorithm to obtain a trained GRU model;
constructing an input vector by using battery capacity data in the kth charging and discharging period of the lithium ion battery to be predicted;
substituting the input vector into the GRU model obtained in the step three to obtain a predicted value of the battery capacity of the lithium ion battery in the (k + 1) th charging and discharging period to be predicted;
step six, judging whether the predicted value obtained in the step five is less than or equal to 80% of the rated capacity of the lithium ion battery to be predicted or not, if yes, executing the step seven, otherwise, enabling k to be k +1, and then returning to the step four;
step seven, taking the difference value between the charge-discharge period corresponding to the predicted value and the actual charge-discharge period of the battery as the residual life of the lithium ion battery to be predicted;
the training set of the GRU model described in step two is represented as [ xtrain, ytrain ], where:
Figure FDA0003498204670000011
in the above formula, CaplThe capacity of the first battery is shown,
Figure FDA0003498204670000012
represents the number of charge and discharge cycles corresponding to the condition that the capacity of the first battery is degraded to 80% of the rated capacity, n represents the length of the capacity test sequence, both l and n represent positive integers, and
Figure FDA0003498204670000013
GRU model includes an update gate ztAnd a reset gate rtLet the t-th column data in xtrain be xtThen, the internal propagation formula of the GRU model is:
zt＝σ(Wz·[ht-1,xt]) (1)
rt＝σ(Wr·[ht-1,xt]) (2)
Figure FDA0003498204670000021
Figure FDA0003498204670000022
yt＝σ(Wo·ht) (5)
wherein the operation symbol portion is]Representing the connection of the two vector endings, the product of the matrix,
Figure FDA0003498204670000023
Of the parameter part
Figure FDA0003498204670000025
Network parameters representing the GRU model,
Figure FDA0003498204670000024
And htAll represent intermediate variables, ytAn estimate representing a GRU model;
the input vector described in step four is expressed as follows:
{Captest(k-n+1),Captest(k-n+2),…,Captest(k)} (6)
Captest(k) and the observed value represents the real capacity of the lithium ion battery in the k-th cycle.
CN201910502496.5A 2019-06-11 2019-06-11 Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion Active CN110222431B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910502496.5A CN110222431B (en) 2019-06-11 2019-06-11 Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910502496.5A CN110222431B (en) 2019-06-11 2019-06-11 Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion

Publications (2)

Publication Number Publication Date
CN110222431A CN110222431A (en) 2019-09-10
CN110222431B true CN110222431B (en) 2022-04-12

Family

ID=67816575

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910502496.5A Active CN110222431B (en) 2019-06-11 2019-06-11 Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion

Country Status (1)

Country Link
CN (1) CN110222431B (en)

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110954132B (en) * 2019-10-31 2023-06-09 太原理工大学 GRNN-assisted self-adaptive Kalman filtering navigation fault identification method
CN111289250A (en) * 2020-02-24 2020-06-16 湖南大学 Method for predicting residual service life of rolling bearing of servo motor
CN111595583B (en) * 2020-05-29 2022-02-01 重庆大学 GAU neural network-based bearing residual life prediction method
CN112098845B (en) * 2020-08-17 2021-08-31 四川大学 Lithium battery state estimation method for distributed energy storage system
CN112098873B (en) * 2020-08-17 2021-06-01 四川大学 Lithium battery health state estimation method based on charging voltage curve geometric characteristics
CN112098874B (en) * 2020-08-21 2023-09-22 杭州电子科技大学 Lithium ion battery electric quantity prediction method considering aging condition
CN112578419B (en) * 2020-11-24 2023-12-12 南京邮电大学 GPS data reconstruction method based on GRU network and Kalman filtering
CN112782594B (en) * 2020-12-24 2022-09-20 杭州电子科技大学 Method for estimating SOC (state of charge) of lithium battery by data-driven algorithm considering internal resistance
CN113009349B (en) * 2021-04-09 2024-01-05 哈尔滨工业大学 Lithium ion battery health state diagnosis method based on deep learning model
CN112986827B (en) * 2021-04-12 2022-06-03 山东凯格瑞森能源科技有限公司 Fuel cell residual life prediction method based on deep learning
CN114861527A (en) * 2022-04-15 2022-08-05 南京工业大学 Lithium battery life prediction method based on time series characteristics
CN114932582B (en) * 2022-06-16 2024-01-23 上海交通大学 Robot small probability failure prediction method based on Bi-GRU self-encoder
CN115267558A (en) * 2022-09-30 2022-11-01 深圳先进技术研究院 Service life prediction and state estimation method of bidirectional lithium ion battery
CN115407211B (en) * 2022-11-01 2023-01-31 北京航空航天大学 Online prediction method and system for health state of lithium battery of electric vehicle
CN115616415B (en) * 2022-12-06 2023-04-07 北京志翔科技股份有限公司 Method, device and equipment for evaluating state of battery pack and storage medium

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103778280A (en) * 2014-01-08 2014-05-07 上海交通大学 Method for remain useful life prognostic of lithium ion battery with model active updating strategy
CN104459560A (en) * 2014-12-29 2015-03-25 合肥工业大学 Method for predicting remaining service life of lithium battery based on wavelet denoising and relevant vector machine
CN104778337A (en) * 2015-04-30 2015-07-15 北京航空航天大学 Method for predicting remaining service life of lithium battery based on FPCA (functional principal component analysis) and Bayesian updating
KR20150093057A (en) * 2014-02-06 2015-08-17 삼성에스디아이 주식회사 Electrolyte for rechargeable lithium battery and rechargeable lithium battery comprising same
CN106093783A (en) * 2016-06-03 2016-11-09 哈尔滨工业大学 The battery SOC method of estimation that Kalman filtering merges with data-driven
CN106908736A (en) * 2017-03-17 2017-06-30 哈尔滨工业大学 Based on the lithium battery method for predicting residual useful life that depth confidence net and Method Using Relevance Vector Machine are merged

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102788955B (en) * 2012-07-17 2015-02-11 哈尔滨工业大学 Remaining lifetime prediction method of ESN (echo state network) turbine generator classification submodel based on Kalman filtering
CN103389471B (en) * 2013-07-25 2015-12-09 哈尔滨工业大学 A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section
CN103399280B (en) * 2013-08-01 2015-08-05 哈尔滨工业大学 Based on the cycle life of lithium ion battery Forecasting Methodology of NSDP-AR model

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103778280A (en) * 2014-01-08 2014-05-07 上海交通大学 Method for remain useful life prognostic of lithium ion battery with model active updating strategy
KR20150093057A (en) * 2014-02-06 2015-08-17 삼성에스디아이 주식회사 Electrolyte for rechargeable lithium battery and rechargeable lithium battery comprising same
CN104459560A (en) * 2014-12-29 2015-03-25 合肥工业大学 Method for predicting remaining service life of lithium battery based on wavelet denoising and relevant vector machine
CN104778337A (en) * 2015-04-30 2015-07-15 北京航空航天大学 Method for predicting remaining service life of lithium battery based on FPCA (functional principal component analysis) and Bayesian updating
CN106093783A (en) * 2016-06-03 2016-11-09 哈尔滨工业大学 The battery SOC method of estimation that Kalman filtering merges with data-driven
CN106908736A (en) * 2017-03-17 2017-06-30 哈尔滨工业大学 Based on the lithium battery method for predicting residual useful life that depth confidence net and Method Using Relevance Vector Machine are merged

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Power-Line Interference Suppression in Electrocardiogram Using Recurrent Neural;Yue Qiu等;《网页在线公开：https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8302231》;20180227;第1-6页 *
基于自适应神经网络的电池寿命退化的预测;史建平等;《电脑技术》;20181121;第42卷(第10期);第1488-1490页 *
改进卡尔曼滤波的融合型锂离子电池SOC估计方法;赵天意等;《仪器仪表学报》;20161018;第37卷(第7期);第1441-1448页 *

Also Published As

Publication number Publication date
CN110222431A (en) 2019-09-10

Similar Documents

Publication Publication Date Title
CN110222431B (en) Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion
CN110568361B (en) Method for predicting health state of power battery
CN108896914B (en) Gradient lifting tree modeling and prediction method for health condition of lithium battery
CN112782591B (en) Lithium battery SOH long-term prediction method based on multi-battery data fusion
CN110568359A (en) lithium battery residual life prediction method
CN111999648A (en) Lithium battery residual life prediction method based on long-term and short-term memory network
CN103399280B (en) Based on the cycle life of lithium ion battery Forecasting Methodology of NSDP-AR model
CN108009585B (en) Lead-acid battery health state prediction method based on local information fusion
CN114861527A (en) Lithium battery life prediction method based on time series characteristics
CN113917334B (en) Battery health state estimation method based on evolution LSTM self-encoder
CN114636932A (en) Method and system for predicting remaining service life of battery
CN111459144A (en) Airplane flight control system fault prediction method based on deep cycle neural network
AU2021101964A4 (en) Artificial intelligence based smart electric vehicle battery management system
CN113391211B (en) Method for predicting remaining life of lithium battery under small sample condition
CN110516813A (en) A method of batteries of electric automobile RDR prediction is carried out based on big data machine learning
CN111709577B (en) RUL prediction method based on long-range correlation GAN-LSTM
CN115453399A (en) Battery pack SOH estimation method considering inconsistency
CN116840720A (en) Fuel cell remaining life prediction method
CN116047300A (en) Controller for predicting characteristic parameters of battery and method thereof
CN115718263A (en) Attention-based lithium ion battery calendar aging prediction model and method
CN115236522A (en) End-to-end capacity estimation method of energy storage battery based on hybrid deep neural network
CN113791351B (en) Lithium battery life prediction method based on transfer learning and difference probability distribution
CN113657030A (en) Method for predicting remaining service life of lithium battery based on Gaussian process regression
CN117289167A (en) Battery remaining life prediction method, device and medium based on multiple neural network
CN116449244A (en) Lithium battery aging track early prediction method and system based on single discharge curve

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