CN116400224A - Battery remaining service life prediction method based on working temperature correction - Google Patents
Battery remaining service life prediction method based on working temperature correction Download PDFInfo
- Publication number
- CN116400224A CN116400224A CN202310319281.6A CN202310319281A CN116400224A CN 116400224 A CN116400224 A CN 116400224A CN 202310319281 A CN202310319281 A CN 202310319281A CN 116400224 A CN116400224 A CN 116400224A
- Authority
- CN
- China
- Prior art keywords
- battery
- capacity
- service life
- aging
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000012937 correction Methods 0.000 title claims abstract description 26
- 230000032683 aging Effects 0.000 claims abstract description 74
- 238000013528 artificial neural network Methods 0.000 claims abstract description 24
- 239000011159 matrix material Substances 0.000 claims abstract description 24
- 230000001133 acceleration Effects 0.000 claims abstract description 17
- 238000005457 optimization Methods 0.000 claims abstract description 13
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 238000012360 testing method Methods 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract description 6
- 238000006243 chemical reaction Methods 0.000 claims description 37
- 230000006870 function Effects 0.000 claims description 25
- 230000004913 activation Effects 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 11
- 230000015654 memory Effects 0.000 claims description 10
- 238000012549 training Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 230000015556 catabolic process Effects 0.000 claims description 5
- 238000006731 degradation reaction Methods 0.000 claims description 5
- GELKBWJHTRAYNV-UHFFFAOYSA-K lithium iron phosphate Chemical compound [Li+].[Fe+2].[O-]P([O-])([O-])=O GELKBWJHTRAYNV-UHFFFAOYSA-K 0.000 claims description 5
- 230000001105 regulatory effect Effects 0.000 claims description 5
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 claims description 3
- 238000013459 approach Methods 0.000 claims description 3
- 239000010439 graphite Substances 0.000 claims description 3
- 229910002804 graphite Inorganic materials 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000005562 fading Methods 0.000 claims 1
- 238000011161 development Methods 0.000 abstract description 2
- 230000018109 developmental process Effects 0.000 abstract description 2
- 230000006403 short-term memory Effects 0.000 abstract description 2
- 230000002194 synthesizing effect Effects 0.000 abstract description 2
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 11
- 229910001416 lithium ion Inorganic materials 0.000 description 11
- 210000004027 cell Anatomy 0.000 description 8
- 230000000694 effects Effects 0.000 description 5
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 4
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 208000019901 Anxiety disease Diseases 0.000 description 2
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 description 2
- 101000872083 Danio rerio Delta-like protein C Proteins 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 2
- 230000036506 anxiety Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000004880 explosion Methods 0.000 description 2
- 230000003834 intracellular effect Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 description 1
- 239000011149 active material Substances 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229910002092 carbon dioxide Inorganic materials 0.000 description 1
- 239000001569 carbon dioxide Substances 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 230000001351 cycling effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 229910052744 lithium Inorganic materials 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000002269 spontaneous effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
Abstract
The invention discloses a battery residual service life prediction method based on working temperature correction, which comprises the steps of simulating the actual running condition of a battery and performing an accelerated aging test by setting a specific charge-discharge strategy, and collecting charge-discharge cycle data for model development and use; the method comprises the steps of performing characteristic extraction on voltage, current and capacity data under each cycle through a multi-layer neural network, processing the data into a sliding matrix with 50 cycles as intervals, and preliminarily predicting the capacity and the residual service life of a battery through a multi-layer Long short-term memory (LSTM); deducing a large amount of accelerated ageing data and performing super-parameter optimization through a big data intelligent algorithm to obtain an Arrhenius RUL prediction semi-empirical model corrected by temperature; and (3) characterizing the aging difference caused by different works by introducing an aging acceleration factor, and synthesizing the preliminary prediction result and the corrected aging deviation to obtain the final accurate prediction of the residual service life of the battery.
Description
Technical Field
The invention relates to the field of lithium ion battery capacity detection, in particular to a battery residual service life prediction method based on working temperature correction.
Background
Lithium ion batteries have become an integral part of our everyday electronic devices and the most advanced electronic mobile devices because of their high energy density and long life cycle. With the increase of global carbon dioxide emission and the continuous development of lithium ion batteries, the electrification of the automobile industry is just as important. However, as the charge and discharge cycles of the battery are increased, the aging and capacity fade phenomena of the lithium ion battery inevitably occur, and the internal resistance increases. Admittedly, the lithium ion battery brings convenience for application in various fields and has potential safety hazards, such as mobile phone battery explosion, lithium battery charger ignition, tesla vehicle spontaneous combustion and the like. The commercialization and popularization of the electric automobile have to put higher requirements on the reliability and safety of the lithium ion battery, therefore, a prediction model is built for the residual service life of the battery system under different use states, reliable basis is provided for the fault prediction and health management of the battery management system, and the method has very important research significance and practical significance.
However, the influence factor of capacity fade of the battery is very complex. Different battery material systems, different operating environment temperatures, different cycling conditions all result in greatly different battery aging rates and capacity manifestations. Particularly for electric vehicles, driving safety and mileage anxiety have been the major concerns. The electric vehicles of the same model have huge performance differences under different environmental temperatures, and the mileage of many electric vehicles is greatly reduced after the electric vehicles travel to the north. This is not only because the low temperature affects the activity of the active material in the battery, but also because of the power rise and the increase in part of the mechanical energy loss of the vehicle air conditioner. When the working temperature of the battery changes, many existing prediction models are difficult to adjust the expansion pertinently, and accurate prediction results are contributed. The accurate residual service life prediction at different working temperatures has important significance for reducing potential safety hazards of vehicles, relieving mileage anxiety and enhancing the robustness of an automobile battery management system.
In summary, the current lithium ion battery residual service life prediction model does not fully consider the service life aging difference caused by the battery operating temperature change, and does not explicitly explore the dependency relationship between the operating temperature and the battery capacity attenuation. Thus, the model does not respond sensitively to changes in operating temperature. In this regard, we propose a battery remaining life prediction based on operating temperature correction.
Disclosure of Invention
The invention provides a battery remaining service life prediction method based on working temperature correction, which aims to make up the defects of the prior art, comprehensively calculate the service life aging factor caused by the battery temperature change, study the dependency relationship between the working temperature and the battery capacity attenuation, wake up the response of a model to the working temperature change, and provide the battery remaining service life prediction method based on the working temperature correction.
In order to solve the technical problems and achieve the purposes, the technical scheme provided by the invention is a battery residual service life prediction method based on working temperature correction, comprising the following steps:
s1: setting a special charge-discharge strategy, simulating the actual running condition of a battery and accelerating the aging of the battery, and recording the data of a charge process and the aging acceleration experimental data;
s2: noise reduction and abnormal value removal are respectively carried out on voltage, current, temperature and historical capacity data in the charging process, the data are regulated into a data structure with the same shape, a time sequence feature extraction module is input, and compact features of variables under different cycle numbers are extracted;
s3: inputting the compact characteristics into a multi-layer long-short-term memory neural network prediction model, performing super-parameter space search through a Bayes optimization algorithm, and determining the optimal super-parameters to obtain a preliminary capacity attenuation curve and a residual service life prediction value;
s4: inputting actual working temperature data of a battery into an Arrhenius temperature correction model, searching optimal semi-empirical model super-parameters through a Bayesian optimization algorithm, and obtaining aging acceleration factors representing different working temperatures;
s5: and combining the preliminary predicted value of the residual service life and the aging acceleration factor to obtain the corrected residual service life prediction.
Further, the step S1 specifically includes:
s101: taking a lithium iron phosphate (lithium iron phosphate, LFP)/graphite battery, wherein the nominal capacity is 1.1Ah, and the nominal voltage is 3.3V;
s102: the battery is charged to the SOC value of the first stage under the high multiplying power C1 in a constant current mode, then is charged to 80% of rated capacity under the second multiplying power C2 in a constant current mode, then is charged to 3.6V under the condition of 1C, and finally is switched to be charged to the constant voltage of 3.6V until the current is smaller than 0.05C; the temperature is 20 ℃ at intervals, and the temperature range is from 0 ℃ to 40 ℃;
s103: according to 8: the ratio of 2 is randomly divided into training and test sets.
Further, the step S102, the battery is cycled to failure in the 48-pass Arbin LBT charge-discharge station and oven.
Further, the step S2 specifically includes:
s201: removing noise of the data through data smoothing and abnormal value removal, and regulating the data into the same matrix shape through an Axma interpolation method, and providing the matrix shape for proper input of a time sequence feature extraction module;
s202: the input data is passed through a multi-layer neural network to generate compact timing characteristics.
Further, the input data generates compact time sequence characteristics through a multi-layer neural network, and specifically comprises the following steps: the input matrix is given with connection weight, then is used as output after being subjected to an activation function, the value of the connection weight determines the importance of the data quantity transmitted by the corresponding neuron, the activation function determines whether the state of the corresponding hidden layer neuron is activated, and the key equation of the neural network y is as follows:
wherein n is the number of neuronal cells, x i Is input to the neural network, w i For the weight, θ is the bias, f (·) is the activation function, here the hyperbolic tangent function f (x):
where x generally refers to the input of the active layer function. The training process enables the model output to approach the optimal compact feature by iteratively adjusting the connection weights.
Further, the step S3 specifically includes:
s301, processing the compact features under each cycle into a slipping matrix with 50 cycles as intervals, and providing matrix input from sequence to sequence of battery life prediction, namely, predicting a time sequence consisting of a plurality of capacity estimation values at future time by a time sequence consisting of a plurality of capacity degradation data at historical time, wherein the obtained slipping matrix is as follows:
in the slipping matrix here, the slipping matrix represents vectors of input and output, respectively, and the sequence length of the input and output is 1000 and 50, respectively, meaning that the capacity performance of 1001 to 1050 cycles of battery in the future will be predicted from 1000 pieces of history capacity information. Every 50 steps of prediction, 1000 cycles of historical capacity before the prediction starting point are used for iteratively predicting the capacity performance of the next 50 cycles. Such predictive methods have fewer iterations and less error accumulation, i.e., a sequence-to-sequence model;
s302, a deep long-short-term memory neural network processes a slipping matrix, predicts capacity under each cycle after a prediction starting point, searches the optimal network node number, network layer number, activation function and learning rate through a Bayesian optimization algorithm, and the key equations of a gate function and a processing process in a single long-short-term memory neural network cell are as follows:
f t ＝σ(W xf *X t +W hf *h t-1 +b f )；
i t ＝σ(W Xi *X t +W hg *h t-1 +b i )；
g t ＝tanh(W Xg *X t +W hg *h t-1 +b g )；
o t ＝σ(W Xo *X t +W ho *h t-1 +b o )；
C t ＝f t ⊙C t-1 +i t ⊙g t ；
h t ＝o t ⊙tanh(C t )；
wherein f t 、i t And o t Respectively representing forgetting gate output, input gate output and output gate output of LSTM nerve cells at t time, wherein the three gates can respectively control the information transmission process in different modes; g t Representing cell candidate memory at time t; c (C) t-1 Indicating the intracellular state at the previous time t-1; c (C) t An intermediate state representing the state of the cell interior at a current time t, the state being varied by a nonlinear function; h is a t For the hidden layer output at time t, xt represents the slip matrix input at time t, W X～ And W is h～ Is the weight of the neural network, namely f, i, g and o are respectively represented by the weight coefficients of corresponding gate functions, and b is the same ～ Representing the bias value of the corresponding gate function. The symbol "×" represents the product operation, ".
Further, the step S4 specifically includes:
s401: accelerated aging test in batteriesIn the experiment, the temperature dependence of the Arrhenius model for representing the residual service life is introduced, the battery capacity attenuation is approximated to a chemical reaction, the Arrhenius equation is used for describing the influence of temperature on the aging reaction and simulating the aging degree of the battery capacity, a new constant b is introduced for representing the future relation between the aging rate and the reaction rate constant K, and the aging rate D at the battery working temperature T is introduced T The following are provided:
D T ＝b·K
at this time, an expression combining the reaction rate and the aging rate can be obtained:
wherein A is the rate constant of the capacity aging reaction, and the aging reaction rate constant A at the temperature T is set to be convenient for subsequent operation T =b·a, and the term-shifting process, can yield:
the derived formula shows that the aging rate D is as long as the aging degradation mechanism of the battery does not exceed a specific temperature range T Is a homogeneous relationship with temperature T and can be characterized by an Arrhenius model, lnDT and
is of slope +.>
Is a linear relationship of (c).
S402: taking a battery with an operating temperature of 40 ℃ as a life reference, and recording the operating temperature at the moment as a reference temperature T 0 The aging rate at this temperature is recorded as
Introducing an aging process at 20 degree centigrade as intervalThe speed factor epsilon, which corrects the remaining service life of the battery at different operating temperatures, is defined as:
at this time, the reaction rate D T The expression of (2) is substituted into the above expression:
wherein E is a R is the gas constant of the capacity aging reaction, T is the working temperature of the capacity aging reaction, and the aging acceleration factor epsilon is used for correcting the service life deviation of the battery caused by different working temperatures;
s403: correcting the capacity fade curve obtained in the step S3 based on the operating temperature, wherein the larger the battery aging rate is, the larger the offset of the capacity fade of the battery is, and the offset is affected by the number of cycles N, so that the available capacity offset is:
ΔC T ＝C 0 ·ε·N n
wherein DeltaC T For the N-th cycle capacity offset at the working temperature T0, C 0 The predicted capacity of the nth turn at the reference operating temperature; n is a model super parameter, and is determined by a super parameter optimization algorithm.
Further, the aging acceleration factor epsilon is used for correcting the aging reaction rate at different working temperatures so as to correct the residual service life of the battery at different working temperatures.
Further, the reaction rate K of the aging reaction is expressed as:
wherein A is the rate constant of the capacity aging reaction, E a Activation energy for Capacity aging reactionR is the gas constant of the capacity aging reaction, T is the working temperature of the capacity aging reaction, and the higher the working temperature of the battery is, the faster the rate of the aging reaction is.
Further, the step S5 specifically includes:
calculating the deviation of the predicted capacity curve at different temperatures to obtain corrected battery capacity until the end of the battery life is as follows:
C T，i ＝C 0，i -ΔC T，i
wherein i=1, 2,.. EOL ，N EOL C for the number of charge and discharge cycles the battery undergoes to end of life 0，i For the capacity of the ith turn at the reference operating temperature ΔC T，i Is the capacity deviation of the ith circle under the real-time working temperature. At this time, the corrective predictive value of the remaining service life can be expressed as:
N RUL ＝N EOL -N i
wherein N is RUL Represents charge-discharge cycles undergone from the predicted start point to the end of life, N i The cycle life is a deviation value due to different temperatures.
The beneficial technical effects of the invention are at least as follows:
(1) The method for predicting the residual service life of the lithium ion battery extracts the characteristics from the charging process data of the battery, can avoid the influence of random workload, and has strong practicability and stability.
(2) According to the method for predicting the residual service life of the lithium ion battery, the RUL is predicted by using the long-short-term memory neural network according to the compact charging characteristics under each cycle, so that the problems of traditional long-sequence gradient explosion and long-term dependence are effectively alleviated.
(3) According to the method for predicting the residual service life of the lithium ion battery, the dependence relationship between the working temperature of the battery and the aging rate and the cycle number of the battery is deduced for the first time based on an Arrhenius equation, and an accurate semi-empirical model is built to correct the deviation of the temperature on RUL prediction.
(4) According to the method for predicting the residual service life of the lithium ion battery, the aging attenuation of the battery is taken as an integral reaction, a complete reaction dynamics equation is established, the aging acceleration factor epsilon is used for correcting the service life of the battery at different working temperatures, and the response sensitivity of the model to the temperature is improved.
Drawings
The invention will be further described with reference to the accompanying drawings, in which embodiments do not constitute any limitation of the invention, and other drawings can be obtained by one of ordinary skill in the art without inventive effort from the following drawings.
FIG. 1 is a flow chart showing the steps of a method for predicting the remaining life of a battery based on operating temperature correction according to the present invention;
FIG. 2 is a graph of capacity fade for 124 battery data sets used in an embodiment of the present invention;
FIG. 3 is a schematic diagram of a neural network according to an embodiment of the present invention;
FIG. 4 is a schematic diagram showing the structure of LSTM cells according to the embodiment of the invention.
Fig. 5 is a schematic diagram of the effect and accuracy of the correction based on the operating temperature when performing RUL prediction according to the embodiment of the present invention.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the present patent;
it will be appreciated by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical scheme of the invention is further described below with reference to the accompanying drawings and examples.
As shown in fig. 1, a method for predicting remaining service life of a battery based on working temperature correction specifically includes the following steps:
Specifically, in step 1.1, lithium iron phosphate (lithium iron phosphate, LFP)/graphite battery was taken, the nominal capacity was 1.1Ah, and the rated voltage was 3.3V.
In step 1.2, the battery is charged to the SOC value of the first stage under a Constant current (CC 1) at a high rate (current) C1, then charged to 80% of the rated capacity under a Constant current (CC 2) at a second rate C2, then charged to 3.6V at a current (CC 3), and finally switched to a Constant Voltage (CV) of 3.6V to a current of less than 0.05C. The working temperature is at intervals of 20 ℃ and ranges from 0 ℃ to 40 ℃. The battery was cycled to failure in a 48-pass Arbin LBT charge-discharge workstation and oven. As shown in fig. 2, a capacity fade plot for 124 battery data sets is given.
Step 1.3, according to 8: the ratio of 2 is randomly divided into training and test sets.
Step 2, respectively carrying out noise reduction and abnormal value removal on the voltage, current, temperature and historical capacity data in the charging process, regulating the data into a data structure with the same shape, inputting the data structure into a time sequence feature extraction module, and extracting compact features of each variable under different cycle numbers;
specifically, in step 2.1, noise of the data is removed by smoothing the data and removing outliers, etc., and the data is normalized into the same matrix shape by an aclma (akima) interpolation method to be provided to a proper input of the timing feature extraction module.
And 2.2, generating compact time sequence characteristics of the input data through the multi-layer neural network. The input matrix is given connection weight and then is subjected to an activation function to be used as output. The value of the connection weight determines the importance of the amount of data transmitted by the corresponding neuron, and the activation function determines whether the state of the corresponding hidden layer neuron is activated. The key equations for the neural network y are as follows:
as shown in fig. 3, a neuronal network structure is given.
Wherein n is the number of neuronal cells, x i Is input to the neural network, w i For the weight, θ is the bias, f (·) is the activation function, here the hyperbolic tangent function:
the training process enables the model output to approach the optimal compact feature by iteratively adjusting the connection weights.
And step 3, inputting the compact features into a multi-layer long-short-term memory neural network prediction model, and performing super-parameter space search through a Bayesian optimization algorithm to determine the optimal super-parameters. And the initial capacity attenuation curve and the residual service life predicted value are obtained by predicting by a sequence-to-sequence method, so that the accumulated error of the traditional sequence prediction is reduced.
Specifically, step 3.1, the compact features under each cycle are processed into a slipping matrix at intervals of 50 cycles to provide a suitable matrix input for achieving a sequence-to-sequence battery life prediction, i.e. a time sequence consisting of a number of capacity degradation data at historical moments to predict a time sequence consisting of a number of capacity estimates at future moments, the resulting slipping matrix being as follows:
in the slipping matrix here, the sequence length of the input and output is 1000 and 50, respectively, meaning that the capacity behavior of 1001 to 1050 battery turns in the future will be predicted from 1000 pieces of history capacity information. Every 50 steps of prediction, 1000 cycles of historical capacity before the prediction starting point are used for iteratively predicting the capacity performance of the next 50 cycles. Such a prediction method is iterated less often and the error accumulation is small, i.e. the sequence-to-sequence model.
And 3.2, processing the sliding matrixes by a deep long-short term memory neural network to realize the prediction of capacity under each cycle after the prediction starting point. And searching the optimal network node number, network layer number, activation function and learning rate through a Bayesian optimization algorithm. The key equations for the gate function and processing procedure within a single long and short term memory neural network cell are as follows:
f t ＝σ(W xf *X t +W hf *h t-1 +b f )；
i t ＝σ(W xi* X t +W hg *h t-1 +b i )；
g t ＝tanh(W Xg *X t +W hg *h t-1 +b g )；
o t ＝σ(W xo *X t +W ho *h t-1 +b o )；
C t ＝f t ⊙C t-1 +i t ⊙ g t ；
h t ＝o t ⊙ tanh(C t )；
wherein f t 、i t And o t Respectively representing forgetting gate output, input gate output and output gate output of LSTM nerve cells at t time, wherein the three gates can respectively control the information transmission process in different modes; g t Representing cell candidate memory at time t; c (C) t-1 Indicating the intracellular state at the previous time t-1; c (C) t An intermediate state representing the state of the cell interior at a current time t, the state being varied by a nonlinear function; h is a t For hidden layer output at time t, X t Represents the slip matrix input at time t, W X～ And W is h～ Is the weight of the neural network, namely f, i, g and o are respectively represented by the weight coefficients of corresponding gate functions, and b is the same ～ Representing the bias value of the corresponding gate function. The symbol "×" represents the product operation, ".
As shown in FIG. 4, the structure of LSTM cells is schematically shown.
Step 4, inputting actual working temperature data of the battery into an Arrhenius temperature correction model, searching optimal semi-empirical model hyper-parameters through a Bayesian optimization algorithm, and obtaining aging acceleration factors representing different working temperatures;
specifically, in step 4.1, an Arrhenius model is introduced to characterize the temperature dependence of the remaining service life in an accelerated aging test of the battery. The battery capacity fade was approximated as a chemical reaction and the effect of temperature on this aging reaction and the degree of aging simulating the battery capacity was described by the arrhenius equation. The chemical reaction rate constant of this reaction can be expressed as:
wherein A, E a R and T are the rate constant, activation energy, gas constant and operating temperature, respectively, of the capacity aging reaction. From this equation, the higher the operating temperature of the battery, the faster the rate of the aging reaction. At this time, a new constant b is introduced to characterize the future relationship between the aging rate and the reaction rate constant K, and the aging rate D at the battery operating temperature T is introduced T The following are provided:
D T ＝b·K
at this time, an expression combining the reaction rate and the aging rate can be obtained:
to facilitate the subsequent operation, let A T =b·a, and the term-shifting process, can yield:
the derived formula shows that the aging rate D is as long as the aging degradation mechanism of the battery does not exceed a specific temperature range T Is a homogeneous relationship with temperature T and can be characterized by an Arrhenius model, lnDT and
is of slope +.>
Is a linear relationship of (c).
Step 4.2, taking the battery with the working temperature of 40 ℃ as a life reference, wherein the working temperature is recorded as a reference temperature T0, and the aging rate at the temperature is recorded as
At 20 degrees celsius as a separation, an aging acceleration factor is introduced that will correct the aging reaction rate at different operating temperatures, thereby correcting the remaining service life of the battery at different operating temperatures, defined as:
at this time, the reaction rate D T The expression of (2) is substituted into the above expression:
the aging acceleration factor can be used for correcting the service life deviation of the battery caused by different working temperatures.
And 4.3, correcting the primary capacity curve obtained in the step three based on the working temperature. The larger the battery aging rate, the larger the offset of the capacity fade of the battery, and the offset is affected by the number of cycles N, so the available capacity offset is:
ΔC T ＝C 0 ·ε·N n
wherein DeltaC T For the N-th cycle capacity offset at the working temperature T0, C 0 The predicted capacity of the nth turn at the reference operating temperature; n is a model super parameter, and is determined by a super parameter optimization algorithm.
And 5, synthesizing the initial predicted value of the RUL and the aging acceleration factor to obtain the corrected RUL prediction.
Specifically, in step 5.1, the deviation of the predicted capacity curve at different temperatures is calculated to obtain the corrected battery capacity until the End of life (EOL) of the battery is as follows:
C T,i ＝C 0,i -ΔC T,i
wherein i=1, 2,.. EOL ，N EOL C for the number of charge and discharge cycles the battery undergoes to end of life 0，i For the capacity of the ith turn at the reference operating temperature ΔC T，i Is the capacity deviation of the ith circle under the real-time working temperature. At this time, the corrective predictive value of the remaining service life can be expressed as:
N RUL ＝N EOL -N i
wherein N is RUL Represents charge-discharge cycles undergone from the predicted start point to the end of life, N i The cycle life is a deviation value due to different temperatures. As shown in fig. 5, the life prediction effect and accuracy of the method disclosed by the invention after working temperature correction are given (two batteries in the experimental data set are taken for display).
Currently, the method disclosed by the invention uses a battery running at 40 ℃ as a reference to perform preliminary prediction on a capacity curve of the battery after a prediction starting point, the average absolute percentage error (Mean Absolute Percentage Error, MAPE) of the prediction is 0.105Ah, and the RUL predicted MAPE is within 20 cycles.
The method successfully establishes an Arrhenius semi-empirical model based on temperature correction through a large amount of data deduction and super-parametric optimization of a large data intelligent algorithm, and when the working temperature of a battery changes, the model method can correct a preliminary life prediction result according to the temperature difference, the corrected capacity prediction MAPE is 0.028Ah, the RUL prediction MAPE is within 5 cycles, and the accuracy is greatly improved.
While embodiments of the invention have been shown and described, it will be understood by those skilled in the art that: many changes, modifications, substitutions and variations may be made to the embodiments without departing from the spirit and principles of the invention, the scope of which is defined by the claims and their equivalents.
Claims (10)
1. The battery remaining service life prediction method based on the working temperature correction is characterized by comprising the following steps of:
s1: setting a special charge-discharge strategy, simulating the actual running condition of a battery and accelerating the aging of the battery, and recording the data of a charge process and the aging acceleration experimental data;
s2: noise reduction and abnormal value removal are respectively carried out on voltage, current, temperature and historical capacity data in the charging process, the data are regulated into a data structure with the same shape, a time sequence feature extraction module is input, and compact features of variables under different cycle numbers are extracted;
s3: inputting the compact characteristics into a multi-layer long-short-term memory neural network prediction model, performing super-parameter space search through a Bayes optimization algorithm, and determining the optimal super-parameters to obtain a preliminary capacity attenuation curve and a residual service life prediction value;
s4: inputting actual working temperature data of a battery into an Arrhenius temperature correction model, searching optimal semi-empirical model super-parameters through a Bayesian optimization algorithm, and obtaining aging acceleration factors representing different working temperatures;
s5: and combining the preliminary predicted value of the residual service life and the aging acceleration factor to obtain the corrected residual service life prediction.
2. The method for predicting remaining service life of a battery based on correction of operating temperature according to claim 1, wherein the step S1 is specifically:
s101: taking a lithium iron phosphate/graphite battery, wherein the nominal capacity is 1.1Ah, and the nominal voltage is 3.3V;
s102: the battery is charged to the SOC value of the first stage under the high multiplying power C1 in a constant current mode, then is charged to 80% of rated capacity under the second multiplying power C2 in a constant current mode, then is charged to 3.6V under the condition of 1C, and finally is switched to be charged to the constant voltage of 3.6V until the current is smaller than 0.05C; the temperature is 20 ℃ at intervals, and the temperature range is from 0 ℃ to 40 ℃;
s103: according to 8: the ratio of 2 is randomly divided into training and test sets.
3. The method for predicting remaining life of a battery based on correction of operating temperature as recited in claim 2, wherein said step S102, said battery is cycled to failure in a 48-pass Arbin LBT charge-discharge workstation and an incubator.
4. The method for predicting remaining service life of a battery based on correction of operating temperature according to claim 1, wherein the step S2 is specifically:
s201: removing noise of the data through data smoothing and abnormal value removal, and regulating the data into the same matrix shape through an Axma interpolation method, and providing the matrix shape for proper input of a time sequence feature extraction module;
s202: the input data is passed through a multi-layer neural network to generate compact timing characteristics.
5. The method for predicting remaining battery life based on operating temperature correction of claim 4, wherein the input data is generated into a compact timing profile via a multi-layer neural network, comprising: the input matrix is given with connection weight, then is used as output after being subjected to an activation function, the value of the connection weight determines the importance of the data quantity transmitted by the corresponding neuron, the activation function determines whether the state of the corresponding hidden layer neuron is activated, and the key equation of the neural network y is as follows:
wherein n is the number of neuronal cells, x i Is input to the neural network, w i For the weight, θ is the bias, f (·) is the activation function, here the hyperbolic tangent function f (x):
where x generally refers to the input of the active layer function. The training process enables the model output to approach the optimal compact feature by iteratively adjusting the connection weights.
6. The method for predicting remaining service life of a battery based on correction of operating temperature according to claim 1, wherein the step S3 is specifically:
s301, processing the compact features under each cycle into a slipping matrix with 50 cycles as intervals, and realizing battery life prediction from sequence to provide matrix input, namely, predicting a time sequence consisting of a plurality of capacity estimation values at future time by a time sequence consisting of a plurality of capacity degradation data at historical time;
s302, the deep long-short-term memory neural network processes the slipping matrix, predicts the capacity under each cycle after predicting the starting point, and searches the optimal network node number, network layer number, activation function and learning rate through a Bayesian optimization algorithm.
7. The method for predicting remaining service life of a battery based on correction of operating temperature according to claim 1, wherein the step S4 is specifically:
s401: in an accelerated aging test of a battery, introducing an Arrhenius model to represent the temperature dependence of the residual service life, approximating the capacity attenuation of the battery to a chemical reaction, and describing the influence of the temperature on the aging reaction and simulating the aging degree of the capacity of the battery by using an Arrhenius equation;
s402: taking a battery with an operating temperature of 40 ℃ as a life reference, wherein the operating temperature is recorded as T 0 Introducing an aging acceleration factor at intervals of 20 ℃ to correct the residual service life of the battery at different working temperatures;
s403: and (3) correcting the capacity fading curve obtained in the step (S3) based on the working temperature.
8. The method for predicting remaining service life of a battery based on correction of operating temperature as set forth in claim 7, wherein the aging acceleration factor is used to correct the aging reaction rate at different operating temperatures and thus correct the remaining service life of the battery at different operating temperatures.
9. The method for predicting remaining life of a battery based on correction of operating temperature as set forth in claim 7, wherein the reaction rate K of the aging reaction is expressed as:
wherein A is the rate constant of the capacity aging reaction, E a The activation energy of the capacity aging reaction is R, the gas constant of the capacity aging reaction is R, and the working temperature of the capacity aging reaction is T.
10. The method for predicting remaining service life of a battery based on correction of operating temperature according to claim 1, wherein the step S5 is specifically:
calculating the deviation of the predicted capacity curve at different temperatures to obtain corrected battery capacity until the End of Life (EOL) of the battery is as follows:
C T,i ＝C 0,i -ΔC T,i
wherein i=1, 2, …, N EOL ，N E0L C for the number of charge and discharge cycles the battery undergoes to end of life 0,i For the capacity of the ith turn at the reference operating temperature ΔC T,i Is the capacity deviation of the ith circle under the real-time working temperature. At this time, the corrective predictive value of the remaining service life can be expressed as:
N RUL ＝N EOL -N i
wherein N is RUL Represents charge-discharge cycles undergone from the predicted start point to the end of life, N i The cycle life is a deviation value due to different temperatures.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310319281.6A CN116400224A (en) | 2023-03-28 | 2023-03-28 | Battery remaining service life prediction method based on working temperature correction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310319281.6A CN116400224A (en) | 2023-03-28 | 2023-03-28 | Battery remaining service life prediction method based on working temperature correction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116400224A true CN116400224A (en) | 2023-07-07 |
Family
ID=87019243
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310319281.6A Pending CN116400224A (en) | 2023-03-28 | 2023-03-28 | Battery remaining service life prediction method based on working temperature correction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116400224A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117526526A (en) * | 2024-01-02 | 2024-02-06 | 深圳市安德普电源科技有限公司 | Battery safety management method, device and equipment |
-
2023
- 2023-03-28 CN CN202310319281.6A patent/CN116400224A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117526526A (en) * | 2024-01-02 | 2024-02-06 | 深圳市安德普电源科技有限公司 | Battery safety management method, device and equipment |
| CN117526526B (en) * | 2024-01-02 | 2024-04-02 | 深圳市安德普电源科技有限公司 | Battery safety management method, device and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dai et al. | A novel estimation method for the state of health of lithium-ion battery using prior knowledge-based neural network and Markov chain | |
| Xu et al. | A novel data-driven method for predicting the circulating capacity of lithium-ion battery under random variable current | |
| Chen et al. | State-of-charge estimation of lithium-ion battery using an improved neural network model and extended Kalman filter | |
| Wang et al. | Perspectives and challenges for future lithium-ion battery control and management | |
| Cui et al. | A dynamic spatial-temporal attention-based GRU model with healthy features for state-of-health estimation of lithium-ion batteries | |
| Che et al. | Semi-supervised self-learning-based lifetime prediction for batteries | |
| Guo et al. | Prediction of remaining useful life and state of health of lithium batteries based on time series feature and Savitzky-Golay filter combined with gated recurrent unit neural network | |
| CN114861527A (en) | Lithium battery life prediction method based on time series characteristics | |
| CN111208438B (en) | Method for cooperatively estimating residual capacity of lithium-ion battery and sensor deviation based on neural network and unscented Kalman filter | |
| CN112215434A (en) | LSTM model generation method, charging duration prediction method and medium | |
| Wang et al. | Health diagnosis for lithium-ion battery by combining partial incremental capacity and deep belief network during insufficient discharge profile | |
| Wei et al. | State of health assessment for echelon utilization batteries based on deep neural network learning with error correction | |
| Li et al. | Lithium-ion battery remaining useful life prognostics using data-driven deep learning algorithm | |
| CN115308606B (en) | Lithium ion battery health state estimation method based on adjacent characteristics | |
| CN116400224A (en) | Battery remaining service life prediction method based on working temperature correction | |
| Cao et al. | A flexible battery capacity estimation method based on partial voltage curves and polynomial fitting | |
| CN115201686A (en) | Lithium ion battery health state assessment method under incomplete charging and discharging data | |
| Hou et al. | The state of health prediction of Li-ion batteries based on an improved extreme learning machine | |
| CN116298936A (en) | Intelligent lithium ion battery health state prediction method in incomplete voltage range | |
| Wen et al. | Lithium battery health state assessment based on vehicle-to-grid (V2G) real-world data and natural gradient boosting model | |
| CN114545279A (en) | Lithium battery health state estimation method based on neural network ordinary differential equation | |
| Hu et al. | State-of-charge estimation for lithium-ion batteries of electric vehicle based on sensor random error compensation | |
| CN118094464A (en) | Lithium battery residual life prediction method integrating projection operator and conditional variation self-encoder | |
| Zhao et al. | State of health estimation of electric vehicle batteries using transformer-based neural network | |
| Fan et al. | An adaptive neuro-fuzzy inference system (ANFIS) based model for the temperature prediction of lithium-ion power batteries |
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 |