CN107544033A - Digital-analog fusion prediction method for remaining service life of lithium ion battery - Google Patents
Digital-analog fusion prediction method for remaining service life of lithium ion battery Download PDFInfo
- Publication number
- CN107544033A CN107544033A CN201710791239.9A CN201710791239A CN107544033A CN 107544033 A CN107544033 A CN 107544033A CN 201710791239 A CN201710791239 A CN 201710791239A CN 107544033 A CN107544033 A CN 107544033A
- Authority
- CN
- China
- Prior art keywords
- lithium battery
- remaining life
- prediction
- high temperature
- life
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
- Tests Of Electric Status Of Batteries (AREA)
Abstract
The invention discloses a digital-analog fusion prediction method for the remaining service life of a lithium battery, which comprises the steps of carrying out model training by using low-temperature capacity degradation data of the lithium battery in a lithium battery historical database, the low-temperature remaining service life of the lithium battery and the low-temperature remaining service life of the lithium battery to obtain a prediction model for the low-temperature remaining service life of the lithium battery; inputting part of low-temperature capacity degradation data of the lithium battery to be tested into the lithium battery low-temperature remaining service life prediction model, and predicting the low-temperature remaining service life of the lithium battery to be tested to obtain a predicted value of the low-temperature remaining service life of the lithium battery; determining a prediction equation of the high-temperature residual service life of the lithium battery to be tested by using the predicted value of the low-temperature residual service life of the lithium battery; and calculating the high-temperature predicted residual service life of the lithium battery by using the determined high-temperature residual service life prediction equation of the lithium battery to be detected.
Description
Technical field
The present invention relates to a kind of lithium battery remaining life Forecasting Methodology merged based on data-driven with model-driven,
It combines data-driven method and the respective advantage of model driven method, completes lithium ion battery in different temperatures
Under remaining life prediction.Suitable for fields such as lithium ion life predictions.
Background technology
Lithium ion battery (is commonly called as lithium battery), is a kind of secondary cell (rechargeable battery), and it relies primarily on lithium ion just
Between pole and negative pole it is mobile come work, have energy density height, open-circuit voltage height, output power, memory-less effect, it is low oneself
Electric discharge, the advantages that operating temperature range is wide, charge/discharge rates are fast, and therefore it is widely used in the electronics such as notebook computer, mobile phone
Product industry；The vehicles industry such as electric bicycle, electric automobile；The national defense and military industry such as Aero-Space；And it is employed
The medical industries such as device are maintained in audiphone, pacemaker and some other non-life.As core component, lithium-ion electric
Loss is all huge caused by pond is on fire.If can before aging of lithium battery if its battery is changed, it should
Prevention that can be to lithium battery accident plays certain positive role, but blindness changes lithium battery ahead of time, will necessarily cause huge
Waste.Therefore, predict that lithium battery remaining life is significant exactly.
The Forecasting Methodology of lithium ion battery remaining life is broadly divided at present based on model-driven, data-driven and
The class of pattern of fusion method three.Method based on model-driven can be modeled to the failure mechanism of lithium battery, can be anti-well
Battery physically and electrically chemical characteristic is reflected, but exists and models the problem of more difficult or model and parameters identification is difficult；Driven based on data
Dynamic method simple practical, it is thus only necessary to which the support of test data and Condition Monitoring Data can be realized as, but be vulnerable to data
Uncertain and imperfection influence, robustness and adaptability are poor.A variety of methods are carried out based on the method for pattern of fusion
It is integrated, the deficiency of single model is made up, the respective advantage of different models is given full play to, more excellent performance can be obtained.But at present
Most of convergence strategy all rests on Decision-level fusion, and simply simple result integrates, it is impossible to plays pattern of fusion method completely
Advantage.The present invention by fused data driving method and model-driven each the advantages of, complete lithium ion battery in not equality of temperature
Remaining life prediction under degree, the life-span management in being actually used whereby to lithium battery aid in.
The content of the invention
It is an object of the invention to provide a kind of lithium ion battery remaining life digital-to-analogue fusion forecasting method, based on data
The method merged with model-driven is driven, the advantages of data-driven method and model driven method is combined, completes lithium electricity
Life prediction of the pond at 25 DEG C, 45 DEG C and 60 DEG C.
According to one embodiment of present invention, a kind of lithium battery remaining life digital-to-analogue fusion forecasting provided by the invention
Method includes：
The longevity is used using lithium battery low temperature capacity degraded data in lithium battery historical data base and lithium battery low temperature are remaining
Life and lithium battery low temperature remaining life carry out model training, obtain lithium battery low temperature remaining life forecast model；
By the way that the part low temperature capacity degraded data of lithium battery to be measured is input to, the lithium battery low temperature is remaining to use the longevity
Order in forecast model, carry out the prediction of lithium battery low temperature remaining life to be measured, obtain lithium battery low temperature remaining life
Predicted value；
Using the lithium battery low temperature remaining life predicted value, determine that lithium battery high temperature remaining life to be measured is pre-
Survey equation；
With identified lithium battery high temperature remaining life predictive equation to be measured, it is remaining to calculate the prediction of lithium battery high temperature
Service life.
According to another embodiment of the invention, a kind of lithium battery remaining life digital-to-analogue fusion provided by the invention is pre-
Survey method includes：
The longevity is used using lithium battery low temperature capacity degraded data in lithium battery historical data base and lithium battery low temperature are remaining
Life and lithium battery low temperature remaining life carry out model training and verification process, obtain lithium battery low temperature remaining life
Forecast model and lithium battery high temperature remaining life amendment predictive equation；
By the way that the part low temperature capacity degraded data of lithium battery to be measured is input to, the lithium battery low temperature is remaining to use the longevity
Order in forecast model, carry out the prediction of lithium battery low temperature remaining life to be measured, obtain lithium battery low temperature remaining life
Predicted value；
Using the lithium battery low temperature remaining life predicted value, determine that lithium battery high temperature remaining life to be measured is pre-
Survey equation；
With identified lithium battery high temperature remaining life predictive equation to be measured, calculating lithium battery high temperature prediction residue makes
It is modified with the life-span, then with the lithium battery high temperature remaining life amendment predictive equation, obtains actual lithium electricity
Pond high temperature remaining life predicted value.
Preferably, it is described to obtain lithium battery low temperature remaining life forecast model and lithium battery high temperature is remaining uses the longevity
Life amendment predictive equation includes：
Utilize linear pre- good to linear prediction effect of lithium battery low temperature capacity degraded data in lithium battery historical data base
The training that model carries out remaining life prediction is surveyed, obtains lithium battery low temperature remaining life forecast model；
Utilize the low remaining life of lithium battery and Ah predicted by lithium battery low temperature remaining life forecast model
Human relations Nice model, determine lithium battery high temperature remaining life predictive equation；
By using lithium battery high temperature real surplus service life value in lithium battery historical data base, remaining by lithium battery high temperature
The lithium battery high temperature of life forecast prediction equation predicts remaining life value and uses the longevity by lithium battery low temperature is remaining
The lithium battery low temperature prediction remaining life value for ordering forecast model prediction carries out verification process, obtains lithium battery high temperature residue
Service life amendment predictive equation.
Preferably, real surplus of the described lithium battery low temperature real surplus service life including 25 DEG C of lithium battery uses the longevity
Life and 45 DEG C of real surplus service life；Described lithium battery low temperature prediction remaining life includes 25 DEG C of lithium battery
Predict remaining life and 45 DEG C of prediction remaining life；Described lithium battery high temperature real surplus service life is
The real surplus service life of 60 DEG C of lithium battery；Described lithium battery high temperature prediction remaining life is 60 DEG C of lithium battery
Predict remaining life.
Preferably, it is described using lithium battery low temperature capacity degraded data in lithium battery historical data base to linear prediction mould
The training that type carries out remaining life predicted value includes：
Utilize 25 DEG C of degradation in capacity data of lithium battery in lithium battery historical data base and 45 DEG C of degradation in capacity data and lithium
The remaining life and 45 DEG C of remaining life that 25 DEG C of battery, moving average model ARIMA is integrated to autoregression and carried out
Training, what is trained is used to be predicted the remaining life of 25 DEG C of lithium battery and 45 DEG C of remaining life
Low temperature remaining life forecast model.
Preferably, the described low remaining life of utilization lithium battery and Arrhenius relationship, determine that lithium battery high temperature remains
Remaining life forecast equation includes：
Using 25 DEG C of remaining life and 45 DEG C of remaining life of lithium battery, determine in Arrhenius relationship
Coefficient；
The fixed Arrhenius relationship of coefficient is used as lithium battery high temperature remaining life predictive equation.
Preferably, using the remaining life of 25 DEG C of lithium battery and the remaining life of 45 DEG C of lithium battery, determine Ah
Coefficient in the model of human relations Nice includes：
It is remaining lifetime value that Arrhenius relationship is simplified into equation ln L=a+b/T, wherein L, and T is temperature, and a is
One coefficient, b are the second coefficient b；
Using the remaining life value of 25 DEG C of lithium battery and the remaining life value of 45 DEG C of lithium battery, ln is solved
The first coefficient a and the second coefficient b in L=a+b/T, so as to obtain the remaining life prediction for 60 DEG C of lithium battery
Arrhenius relationship is simplified to equation, to calculate 60 DEG C of remaining lifes of lithium battery to be measured using equation ln L=a+b/T
Predicted value.
Preferably, the described lithium battery high temperature remaining life amendment predictive equation that obtains includes：
Using a series of lithium battery high temperature actual life values in lithium battery historical data base with being remained by lithium battery high temperature
A series of a series of ratios for lithium battery high temperature prediction remaining life value that remaining life forecast equation calculates, are obtained
A series of a series of ratios of actual life values of lithium battery high temperature high temperature prediction remaining life values a series of with lithium battery
Ratio curve corresponding to value；
It is remaining using a series of prediction for 25 DEG C of the lithium batteries predicted by lithium battery low temperature remaining life forecast model
A series of differences of service life value and a series of prediction remaining life value of 45 DEG C of lithium batteries, obtain a series of lithium electricity
A series of differences of the prediction remaining life value in 25 DEG C of pond and a series of prediction remaining life value of 45 DEG C of lithium batteries
Corresponding difference curve；
According to the ratio curve and the difference curve, a prediction remaining life for making 25 DEG C of lithium battery is fitted
The difference of value and the prediction remaining life value of 45 DEG C of lithium battery is equal to lithium battery high temperature actual life value and lithium battery
The lithium battery high temperature remaining life amendment predictive equation of the ratio of high temperature prediction remaining life value.
Preferably, with the lithium battery high temperature remaining life amendment predictive equation it is modified including：
It is surplus calculating the prediction of lithium battery high temperature with identified lithium battery high temperature remaining life predictive equation to be measured
After remaining service life, it is multiplied by with the result of calculation of the lithium battery high temperature remaining life amendment predictive equation and calculates lithium
Battery high-temperature predicts remaining life value.
Preferably, described lithium battery high temperature remaining life amendment predictive equation is：Ratio=f
(difference)；Wherein, ratio is that lithium battery high temperature actual life value uses the longevity with lithium battery high temperature prediction residue
The ratio of life value；F (difference) is prediction remaining life value and lithium battery 45 DEG C of the variable for 25 DEG C of lithium battery
Predict the fitting function of the difference of remaining life value.
Brief description of the drawings
With reference to accompanying drawing and following detailed description, preceding feature of the invention is more readily understood that, wherein：
Fig. 1 shows lithium battery remaining life Forecasting Methodology flow of the present invention based on digital-to-analogue fusion；
Fig. 2 shows degradation in capacity curve of the lithium battery of the present invention at 25 DEG C, 45 DEG C, 60 DEG C；
Fig. 3 shows the ARIMA life prediction effects at 25 DEG C of A groups battery in lithium battery historical data base of the present invention.
Fig. 4 shows the difference difference of 25 DEG C of bimetries and 45 DEG C of bimetries in historical data of the present invention,
Extrapolated the ratio between life-spans ratio with 60 DEG C of true lifetimes and 60 DEG C of Arrhenius relationships, fit correlation between the two；
Sequence number, symbol, code name are described as follows in figure：
ARIMA：Autoregression integration moving average model (Autoregressive Integrated Moving
Average) (see Fig. 1)
difference:The difference of 25 DEG C of bimetries and 45 DEG C of bimetries；ratio：60 DEG C of true lifetimes with 60 DEG C
The ratio between Arrhenius relationship extrapolation life-span
Embodiment
In the first embodiment of the present invention, a kind of lithium battery remaining life digital-to-analogue fusion forecasting method of the invention
Including：
The longevity is used using lithium battery low temperature capacity degraded data in lithium battery historical data base and lithium battery low temperature are remaining
Life and lithium battery low temperature remaining life carry out model training, obtain lithium battery low temperature remaining life forecast model；
By the way that the part low temperature capacity degraded data of lithium battery to be measured is input to, the lithium battery low temperature is remaining to use the longevity
Order in forecast model, carry out the prediction of lithium battery low temperature remaining life to be measured, obtain lithium battery low temperature remaining life
Predicted value；
Using the lithium battery low temperature remaining life predicted value, determine that lithium battery high temperature remaining life to be measured is pre-
Survey equation；
With identified lithium battery high temperature remaining life predictive equation to be measured, it is remaining to calculate the prediction of lithium battery high temperature
Service life.
In the second embodiment of the present invention, a kind of lithium battery remaining life digital-to-analogue fusion forecasting provided by the invention
Method includes：
The longevity is used using lithium battery low temperature capacity degraded data in lithium battery historical data base and lithium battery low temperature are remaining
Life and lithium battery low temperature remaining life carry out model training and verification process, obtain lithium battery low temperature remaining life
Forecast model and lithium battery high temperature remaining life amendment predictive equation；
By the way that the part low temperature capacity degraded data of lithium battery to be measured is input to, the lithium battery low temperature is remaining to use the longevity
Order in forecast model, carry out the prediction of lithium battery low temperature remaining life to be measured, obtain lithium battery low temperature remaining life
Predicted value；
Using the lithium battery low temperature remaining life predicted value, determine that lithium battery high temperature remaining life to be measured is pre-
Survey equation；
With identified lithium battery high temperature remaining life predictive equation to be measured, calculating lithium battery high temperature prediction residue makes
It is modified with the life-span, then with the lithium battery high temperature remaining life amendment predictive equation, obtains actual lithium electricity
Pond high temperature remaining life predicted value.
The precision for the lithium battery high temperature remaining life predicted value that the method for above-mentioned first embodiment is predicted is less than upper
State the precision for the lithium battery high temperature remaining life predicted value that the method for second embodiment is predicted, thus first embodiment
Method be suitable for the scene that computing resource is less and precision of prediction is not high.
Wherein, it is described to obtain lithium battery low temperature remaining life forecast model and lithium battery high temperature remaining life
Amendment predictive equation includes：
Utilize linear pre- good to linear prediction effect of lithium battery low temperature capacity degraded data in lithium battery historical data base
The training that model carries out remaining life prediction is surveyed, obtains lithium battery low temperature remaining life forecast model；
Utilize the low remaining life of lithium battery and Ah predicted by lithium battery low temperature remaining life forecast model
Human relations Nice model, determine lithium battery high temperature remaining life predictive equation；
By using lithium battery high temperature real surplus service life value in lithium battery historical data base, remaining by lithium battery high temperature
The lithium battery high temperature of life forecast prediction equation predicts remaining life value and uses the longevity by lithium battery low temperature is remaining
The lithium battery low temperature prediction remaining life value for ordering forecast model prediction carries out verification process, obtains lithium battery high temperature residue
Service life amendment predictive equation.
Wherein, described lithium battery low temperature real surplus service life includes the real surplus service life of 25 DEG C of lithium battery
With 45 DEG C of real surplus service life；Described lithium battery low temperature prediction remaining life includes the pre- of 25 DEG C of lithium battery
Survey remaining life and 45 DEG C of prediction remaining life；Described lithium battery high temperature real surplus service life is lithium
The real surplus service life of 60 DEG C of battery；Described lithium battery high temperature prediction remaining life is the pre- of 60 DEG C of lithium battery
Survey remaining life.
Wherein, it is described using lithium battery low temperature capacity degraded data in lithium battery historical data base to linear prediction model
Carrying out the training of remaining life predicted value includes：
Utilize 25 DEG C of degradation in capacity data of lithium battery in lithium battery historical data base and 45 DEG C of degradation in capacity data and lithium
The remaining life and 45 DEG C of remaining life that 25 DEG C of battery, moving average model ARIMA is integrated to autoregression and carried out
Training, what is trained is used to be predicted the remaining life of 25 DEG C of lithium battery and 45 DEG C of remaining life
Low temperature remaining life forecast model.
Wherein, the described low remaining life of utilization lithium battery and Arrhenius relationship, determine lithium battery high temperature residue
Life forecast equation includes：
Using 25 DEG C of remaining life and 45 DEG C of remaining life of lithium battery, determine in Arrhenius relationship
Coefficient；
The fixed Arrhenius relationship of coefficient is used as lithium battery high temperature remaining life predictive equation.
Preferably, using the remaining life of 25 DEG C of lithium battery and the remaining life of 45 DEG C of lithium battery, determine Ah
Coefficient in the model of human relations Nice includes：
It is remaining lifetime value that Arrhenius relationship is simplified into equation ln L=a+b/T, wherein L, and T is temperature, and a is
One coefficient, b are the second coefficient b；
Using the remaining life value of 25 DEG C of lithium battery and the remaining life value of 45 DEG C of lithium battery, ln L are solved
The first coefficient a and the second coefficient b in=a+b/T, so as to obtain the remaining life prediction for 60 DEG C of lithium battery Ah
The model simplification of human relations Nice is into equation, to calculate 60 DEG C of remaining lifes of lithium battery to be measured using equation ln L=a+b/T
Predicted value.
Wherein, the described lithium battery high temperature remaining life amendment predictive equation that obtains includes：
Using a series of lithium battery high temperature actual life values in lithium battery historical data base with being remained by lithium battery high temperature
A series of a series of ratios for lithium battery high temperature prediction remaining life value that remaining life forecast equation calculates, are obtained
A series of a series of ratios of actual life values of lithium battery high temperature high temperature prediction remaining life values a series of with lithium battery
Ratio curve corresponding to value；
It is remaining using a series of prediction for 25 DEG C of the lithium batteries predicted by lithium battery low temperature remaining life forecast model
A series of differences of service life value and a series of prediction remaining life value of 45 DEG C of lithium batteries, obtain a series of lithium electricity
A series of differences of the prediction remaining life value in 25 DEG C of pond and a series of prediction remaining life value of 45 DEG C of lithium batteries
Corresponding difference curve；
According to the ratio curve and the difference curve, a prediction remaining life for making 25 DEG C of lithium battery is fitted
The difference of value and the prediction remaining life value of 45 DEG C of lithium battery is equal to lithium battery high temperature actual life value and lithium battery
The lithium battery high temperature remaining life amendment predictive equation of the ratio of high temperature prediction remaining life value.
Wherein, with the lithium battery high temperature remaining life amendment predictive equation it is modified including：
It is surplus calculating the prediction of lithium battery high temperature with identified lithium battery high temperature remaining life predictive equation to be measured
After remaining service life, it is multiplied by with the result of calculation of the lithium battery high temperature remaining life amendment predictive equation and calculates lithium
Battery high-temperature predicts remaining life value.
Wherein, described lithium battery high temperature remaining life amendment predictive equation is：Ratio=f (difference)；
Wherein, ratio is the ratio of lithium battery high temperature actual life value and lithium battery high temperature prediction remaining life value；f
(difference) it is that variable uses for the prediction remaining life value of 25 DEG C of lithium battery and the prediction residue of 45 DEG C of lithium battery
The fitting function of the difference of life value.
The above method of the present invention is described in detail below in conjunction with the accompanying drawings.
It is remaining that the present invention proposes a kind of lithium battery merged based on data-driven with model-driven for considering different temperatures
Life forecast method, its specific design flow diagram are as shown in Figure 1.
Because the degenerated curve linear degree of lithium battery at different temperatures is different.As shown in Fig. 2 typically in low temperature
Under, linear degree is high, and linear degree is low at high temperature.This just needs to be distinguish between the life prediction under different temperatures, profit
Specific aim solution is carried out with data-driven method and the respective advantage of model driven method.Degradation in capacity at 25 DEG C and 45 DEG C
Curve linear degree is higher, therefore is predicted using data-driven method.Degradation in capacity curve linear degree at 60 DEG C is low, because
This is predicted using Arrhenius relationship.But Arrhenius relationship extrapolation life-span precision itself is not high, it is necessary to modified result
Coefficient carries out auxiliary prediction.
The foundation of modified result coefficient equation：
Step 1：Use the preferable autoregression integration moving average model of linear prediction effect in data-driven method
(Autoregressive Integrated Moving Average, ARIMA) is to 25 of battery in lithium battery historical data base
DEG C and 45 DEG C at remaining life be predicted.
Step 2：Using remaining life predicted value of the gained lithium battery at 25 DEG C and 45 DEG C in step 1 to Allan
Two parameters in the model of Nice are identified, and obtain final Arrhenius relationship.
Step 3：The remaining life at 60 DEG C is carried out using Arrhenius relationship to be predicted, and is obtained lithium battery and is gone through
Remaining life predicted value in history database at 60 DEG C of battery.
Step 4：Repeat step one arrives step 3, completes all 25 DEG C of batteries progress, 45 in lithium battery historical data base
DEG C, the prediction under 60 DEG C of life-spans.Modified result coefficient equation is obtained according to prediction result.
The life prediction of lithium battery to be predicted：
Step 1：To step 1 in battery to be predicted repetition " foundation of modified result coefficient equation " to step 3, and make
The cycle life at 60 DEG C is modified with acquired results correction factor equation in step 4, provide final 25 DEG C, 45 DEG C,
Prediction result at 60 DEG C.
Wherein, " the ARIMA models " in " foundation of modified result coefficient equation " described in step 1, it specifically should
It is as follows with process：
The first step, parameter determine rank.ARIMA (p, d, q) shares auto-regressive parameter p, difference number d, rolling average parameter q
Totally three parameters.The degradation in capacity data of 500 circulations before lithium battery are used as input, first by its difference d time, guarantee difference
Sequence afterwards is stable；Then the auto-regressive equation (Autocorrelation Function, ACF) of sequence after difference is obtained
With inclined auto-regressive equation (Partial Autocorrelation Function, PACF), by observing dragging for ACF and PACF
Tail situation determines auto-regressive parameter p and rolling average parameter q.
Second step, parameter fitting.Pass through regression analysis, the parameter alpha of fitting autoregression (Autoregressive, AR) item
The parameter beta of (Moving Average, MA) item is returned with movement, and it is random to parameter alpha and β progress significance tests and residual error
Property examine.
3rd step, examine and apply.Inspection institute obtains whether ARIMA (p, d, q) model coincide with historical data.And if
In the case that historical data is coincide, application model prediction degradation in capacity situation, remaining life predicted value is finally given.
Wherein, " Arrhenius relationship " in " foundation of modified result coefficient equation " described in step 2, it is established
Process is as follows：
Arrhenius in 1989 is summed up on the basis of research temperature is to acid catalysis sucrose hydrolysis conversion reaction：Certain production
The performance degradation speed of product and the index of activation energy are inversely proportional, and are inversely proportional, can be expressed as with the index of inverse temperature：
In formula, M is the amount of degradation of product characteristic value；, should for deterioration velocity of the temperature at T (thermodynamic temperature)
Deterioration velocity is time t linear function；K is Boltzmann constant, is 8.617 × 10-5eV/℃；T is absolute temperature, etc.
In Celsius temperature plus 273.15；A0It is constant；T is the reaction time；Δ E is that failure mechanism activates energy, unit eV, to same class
The same failure mode of component is constant.
The amount of degradation for making product original state is M1, the corresponding time is t1；The amount of degradation of another state is M2, to it is corresponding when
Between be t2.Then, when T is constant, from t1To t2Accumulation amount of degradation be：
Then have
M2-M1=A0·exp[-ΔE/(kT)]·(t2-t1) (3)
Make t=(t2-t1), then have
As amount of degradation M2Reach some threshold value MpWhen, then it is assumed that the product failure, time difference (t at this momentp-t1) it is exactly to produce
Product are from t1Start the life-span L continued.I.e.
Order：Then have
Ln L=a+b/T (6)
Wherein, a and b is undetermined parameter.Therefore, it is known that any two temperature and at this temperature lithium battery are remaining to use the longevity
Order the life-span, it is possible to substitute into equation (6) and identify the value of two parameters of a, b, and make with the residue under this arbitrary temp of extrapolating
Use the life-span.
Wherein, it is as follows in " the modified result coefficient equation " described in step 4 kind, the method that it is established：
According to the analysis to test data, the difference difference of 25 DEG C of bimetries of discovery and 45 DEG C of bimetries,
Extrapolated the ratio between life-spans ratio with 60 DEG C of true lifetimes and 60 DEG C of Arrhenius relationships, fit correlation between the two be present.Pass through
Fitting to historical data, modified result coefficient can be obtained by fit equation.
Ratio=f (difference) (7)
(1) advantages of the present invention
1. the remaining battery service life method proposed by the present invention based on digital-to-analogue fusion is by data-driven method and model
The advantages of driving method, combines, and the lithium battery pointedly solved at a temperature of 25 DEG C, 45 DEG C, 60 DEG C three kinds is remaining
Life problem, prediction accuracy are high.
In one embodiment of the present of invention, verified using the test data of Ningde epoch new energy Science and Technology Co., Ltd.
Feasibility and the validity (note of the lithium ion battery life-span prediction method proposed：Battery used is a kind of special use in experiment
It is different with the battery that is used in company actual products in the soft-package battery of Design Stage).
According to the difference of the designs such as anode material, electrolyte, the batch lithium battery is divided into A-J totally 10 groups, wherein A-I
Group carries out modified result regression equation recurrence as historical data base, and J groups are verified for final Forecasting Methodology, are made in this example
Data specifying information is as shown in table 1.It is noted that this batch of battery is the test cell of design phase, and non-final electricity
Pond product.
The lithium battery information of table 1
This method comprises the following steps that：
The foundation of modified result coefficient equation：
Step 1：Using 25 DEG C of data of A groups as specific object, its initial data is as shown in figure 4, to before this group of data 500
Individual cycle applications ARIMA (p, d, q), draws model parameter p=1, d=1, q=11, obtains model ARIMA (1,1,11).Through
Parametric regression is crossed, it is -8.865E-5 to obtain constant, the parameter alpha of autoregression item1=0.399, the parameter beta of rolling average item1=
0.767, β11=0.129.Final gained model is as follows：
(Ct-Ct-1)=- 8.865 × 10-5+0.399(Ct-1-Ct-2)+0.767μt-1+0.129μt-11 (8)
Wherein CtIt is the battery capacity under t-th of circulation, μtIt is t-th of numerical value of white noise sequence.It can be seen that the model
It can be good at making prediction to the long-term cycle life of battery, final prediction result is as shown in Figure 3.
Repeat step one, obtain life prediction knot of other batteries at 25 DEG C and 45 DEG C in lithium battery historical data base
Fruit, as shown in table 2.
2 25 DEG C of table and 45 DEG C of bimetries
Step 2：By taking A groups as an example, it is that 3022, A groups, 45 DEG C of bimetries are 2362 to take 25 DEG C of bimetries of A groups, such as table
Shown in lattice 2, by T1=25+273.15；L1=3022 and T2=45+273.15；L1=2362 substitute into equation lnL=a+b/T, calculate
Go out parameter a=4.0939, b=1168.7, it is as follows to finally give Arrhenius relationship：
LnL=4.0939+1168.7/T (9)
Step 3：T=60+273.15 is substituted into equation 9, it is 2002 circulations to obtain the bimetry at 60 DEG C.
Step 4：By A groups in form 2 to 25 DEG C of I groups and 45 DEG C of prediction result combinations, step 2 is carried out as input
And step 3, Arrhenius relationship extrapolation lifetime results of each Battery pack at 60 DEG C are obtained, as shown in table 3.
3 60 DEG C of Arrhenius relationship extrapolation life-spans of table
A groups are into I groups, the difference difference of 25 DEG C of bimetries and 45 DEG C of bimetries, with 60 DEG C of true lifetimes
It has been listed in the ratio between 60 DEG C of Arrhenius relationship extrapolation life-spans ratio in table 4
The modified result equation regression data of table 4
Note：Two columns values correspond by separator of comma in table 4
Regression analysis is carried out to two column datas in form 4, it is as follows to obtain regression equation:
Ratio=2.87e-06difference1.61+0.4629 (10)
The life prediction of lithium battery to be predicted：
Step 1：J Battery packs are predicted, according to form 1, bimetry is 2713 circulations at 25 DEG C,
Bimetry is 1363 circulations at 45 DEG C, and both differences are 1350 circulations.Difference substitution equation (10) is obtained into result to repair
Positive coefficient is 0.7775.According to form 3, it is 859 that both, which are substituted into Arrhenius relationship to obtain extrapolation life-span at 60 DEG C,
Individual circulation.It is 859 × 0.7775 ≈ 668 to be multiplied by the bimetry that modified result coefficient obtained at final 60 DEG C with the extrapolation life-span
Individual circulation.As shown in Table 3, the true lifetime at 60 DEG C of J Battery packs is 703 circulations.Bimetry is with being really life-span difference
Circulated for 668-703=-35, precision of prediction 1- | 668-703 |/703 × 100%=95.02%.
Although the present invention is described in detail above, the invention is not restricted to this, those skilled in the art of the present technique
Various modifications can be carried out according to the principle of the present invention.Therefore, all modifications made according to the principle of the invention, all should be understood
To fall into protection scope of the present invention.
Claims (10)
1. a kind of lithium battery remaining life digital-to-analogue fusion forecasting method, including：
Using lithium battery low temperature capacity degraded data in lithium battery historical data base and lithium battery low temperature remaining life and
Lithium battery low temperature remaining life carries out model training, obtains lithium battery low temperature remaining life forecast model；
It is pre- by the way that the part low temperature capacity degraded data of lithium battery to be measured is input into the lithium battery low temperature remaining life
Survey in model, carry out the prediction of lithium battery low temperature remaining life to be measured, obtain the prediction of lithium battery low temperature remaining life
Value；
Using the lithium battery low temperature remaining life predicted value, lithium battery high temperature remaining life prediction side to be measured is determined
Journey；
With identified lithium battery high temperature remaining life predictive equation to be measured, the remaining use of lithium battery high temperature prediction is calculated
Life-span.
2. a kind of lithium battery remaining life digital-to-analogue fusion forecasting method, including：
Using lithium battery low temperature capacity degraded data in lithium battery historical data base and lithium battery low temperature remaining life and
Lithium battery low temperature remaining life carries out model training and verification process, obtains lithium battery low temperature remaining life prediction mould
Type and lithium battery high temperature remaining life amendment predictive equation；
It is pre- by the way that the part low temperature capacity degraded data of lithium battery to be measured is input into the lithium battery low temperature remaining life
Survey in model, carry out the prediction of lithium battery low temperature remaining life to be measured, obtain the prediction of lithium battery low temperature remaining life
Value；
Using the lithium battery low temperature remaining life predicted value, lithium battery high temperature remaining life prediction side to be measured is determined
Journey；
With identified lithium battery high temperature remaining life predictive equation to be measured, calculate lithium battery high temperature prediction residue and use the longevity
Life, then it is modified with the lithium battery high temperature remaining life amendment predictive equation, it is high to obtain actual lithium battery
Warm remaining life predicted value.
3. according to the method for claim 2, wherein, it is described obtain lithium battery low temperature remaining life forecast model and
Lithium battery high temperature remaining life amendment predictive equation includes：
Utilize the linear prediction mould good to linear prediction effect of lithium battery low temperature capacity degraded data in lithium battery historical data base
Type carries out the training of remaining life prediction, obtains lithium battery low temperature remaining life forecast model；
Utilize the low remaining life of lithium battery and Allan Buddhist nun predicted by lithium battery low temperature remaining life forecast model
This model, determine lithium battery high temperature remaining life predictive equation；
Used by using lithium battery high temperature real surplus service life value in lithium battery historical data base, by lithium battery high temperature is remaining
The lithium battery high temperature prediction remaining life value and pre- by lithium battery low temperature remaining life of life prediction prediction equation
The lithium battery low temperature prediction remaining life value for surveying model prediction carries out verification process, obtains lithium battery high temperature residue and uses the longevity
Life amendment predictive equation.
4. method according to claim 1 or 2, wherein, described lithium battery low temperature real surplus service life includes lithium
The real surplus service life of 25 DEG C of battery and 45 DEG C of real surplus service life；Described lithium battery low temperature prediction residue makes
Include the prediction remaining life and 45 DEG C of prediction remaining life of 25 DEG C of lithium battery with the life-span；Described lithium battery is high
Warm real surplus service life is the real surplus service life of 60 DEG C of lithium battery；Described lithium battery high temperature prediction is remaining to be used
Life-span is the prediction remaining life of 60 DEG C of lithium battery.
5. the method according to right wants 4, wherein, described is moved back using lithium battery low temperature capacity in lithium battery historical data base
Changing training of the data to linear prediction model progress remaining life predicted value includes：
Utilize 25 DEG C of degradation in capacity data of lithium battery in lithium battery historical data base and 45 DEG C of degradation in capacity data and lithium battery
25 DEG C of remaining life and 45 DEG C of remaining life, autoregression integration moving average model ARIMA is trained,
The low temperature for being used to be predicted the remaining life of 25 DEG C of lithium battery and 45 DEG C of remaining life trained
Remaining life forecast model.
6. according to the method for claim 5, wherein, described utilizes the low remaining life of lithium battery and Arrhenius mould
Type, determine that lithium battery high temperature remaining life predictive equation includes：
Using 25 DEG C of remaining life and 45 DEG C of remaining life of lithium battery, determine be in Arrhenius relationship
Number；
The fixed Arrhenius relationship of coefficient is used as lithium battery high temperature remaining life predictive equation.
According to the method for claim 6,7. wherein, 45 DEG C of the remaining life and lithium battery of 25 DEG C of lithium battery are utilized
Remaining life, determine that the coefficient in Arrhenius relationship includes：
It is remaining lifetime value that Arrhenius relationship is simplified into equation ln L=a+b/T, wherein L, and T is temperature, and a is the first system
Number, b is the second coefficient b；
Using the remaining life value of 25 DEG C of lithium battery and the remaining life value of 45 DEG C of lithium battery, ln L=a+ are solved
The first coefficient a and the second coefficient b in b/T, so as to obtain the Allan Buddhist nun of the remaining life prediction for 60 DEG C of lithium battery
This model simplification is into equation, to calculate the prediction of 60 DEG C of remaining lifes of lithium battery to be measured using equation ln L=a+b/T
Value.
8. according to the method for claim 4, wherein, described obtains lithium battery high temperature remaining life amendment prediction side
Journey includes：
Using a series of lithium battery high temperature actual life values in lithium battery historical data base with being made by lithium battery high temperature residue
A series of a series of ratios of the lithium battery high temperature prediction remaining life value calculated with life prediction equation, obtain lithium battery
A series of ratios that a series of high temperature actual lives are worth high temperature prediction remaining life values a series of with lithium battery are corresponding
Ratio curve；
Used using a series of prediction for 25 DEG C of the lithium batteries predicted by lithium battery low temperature remaining life forecast model is remaining
Life value and a series of a series of differences of the prediction remaining life value of 45 DEG C of lithium batteries, obtain a series of 25 DEG C of lithium batteries
Prediction remaining life value and a series of 45 DEG C of lithium batteries prediction remaining life value a series of differences corresponding to
Difference curve；
According to the ratio curve and the difference curve, fitting one make the prediction remaining life value of 25 DEG C of lithium battery with
The difference of the prediction remaining life value of 45 DEG C of lithium battery is equal to lithium battery high temperature actual life value and lithium battery high temperature
Predict the lithium battery high temperature remaining life amendment predictive equation of the ratio of remaining life value.
9. the method according to claim 11, wherein, with the lithium battery high temperature remaining life amendment predictive equation pair
It is modified including：
Make calculating lithium battery high temperature prediction residue with identified lithium battery high temperature remaining life predictive equation to be measured
After the life-span, it is multiplied by with the result of calculation of the lithium battery high temperature remaining life amendment predictive equation and calculates lithium battery height
Temperature prediction remaining life value.
10. the method according to claim 11, wherein, described lithium battery high temperature remaining life amendment predictive equation
For：Ratio=f (difference)；
Wherein, ratio is the ratio of lithium battery high temperature actual life value and lithium battery high temperature prediction remaining life value；
F (difference) is that variable uses for the prediction remaining life value of 25 DEG C of lithium battery and the prediction residue of 45 DEG C of lithium battery
The fitting function of the difference of life value.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2016108011681 | 2016-09-05 | ||
| CN201610801168 | 2016-09-05 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107544033A true CN107544033A (en) | 2018-01-05 |
| CN107544033B CN107544033B (en) | 2020-01-10 |
Family
ID=60959280
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710791239.9A Active CN107544033B (en) | 2016-09-05 | 2017-09-05 | Digital-analog fusion prediction method for remaining service life of lithium ion battery |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107544033B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110058177A (en) * | 2019-05-06 | 2019-07-26 | 奇瑞新能源汽车技术有限公司 | A kind of power battery electricity SOC modification method |
| CN110658462A (en) * | 2019-10-25 | 2020-01-07 | 北京航空航天大学 | Lithium battery online service life prediction method based on data fusion and ARIMA model |
| CN110888077A (en) * | 2019-10-30 | 2020-03-17 | 无锡市产品质量监督检验院 | Accelerated lithium ion battery life evaluation method based on ARIMA time sequence |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0227278A (en) * | 1988-07-15 | 1990-01-30 | Fujitsu Ltd | Battery life checker |
| JP2010117256A (en) * | 2008-11-13 | 2010-05-27 | Toshiba Toko Meter Systems Co Ltd | Battery lifetime prediction apparatus |
| CN102057289A (en) * | 2008-07-02 | 2011-05-11 | 松下电器产业株式会社 | Life span estimation method for a lead acid battery and a power supply system |
| CN102778653A (en) * | 2012-06-20 | 2012-11-14 | 哈尔滨工业大学 | Data-driven lithium ion battery cycle life prediction method based on AR (Autoregressive) model and RPF (Regularized Particle Filtering) algorithm |
| CN103323690A (en) * | 2012-03-23 | 2013-09-25 | 海洋王（东莞）照明科技有限公司 | Service life testing method of lamp charger and service life testing device |
| CN103344923A (en) * | 2013-08-01 | 2013-10-09 | 哈尔滨工业大学 | EKF (extended Kalmar filter)-method and NSDP-AR-model-based method for predicting cycling life of fusion-type lithium ion battery |
| CN103399279A (en) * | 2013-08-01 | 2013-11-20 | 哈尔滨工业大学 | Method for predicting cycle life of fused lithium ion battery based on EKF (Extended Kalman Filter) method and AR (AutoRegressive) model |
| CN103399281A (en) * | 2013-08-01 | 2013-11-20 | 哈尔滨工业大学 | Lithium ion battery cycle life predicating method based on cycle life degeneration stage parameter ND-AR (neutral density-autoregressive) model and EKF (extended Kalman filter) method |
| CN103954913A (en) * | 2014-05-05 | 2014-07-30 | 哈尔滨工业大学深圳研究生院 | Predication method of electric vehicle power battery service life |
| CN104714189A (en) * | 2015-04-02 | 2015-06-17 | 奇瑞汽车股份有限公司 | Method for predicting cycle life of battery pack for electric car |
-
2017
- 2017-09-05 CN CN201710791239.9A patent/CN107544033B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0227278A (en) * | 1988-07-15 | 1990-01-30 | Fujitsu Ltd | Battery life checker |
| CN102057289A (en) * | 2008-07-02 | 2011-05-11 | 松下电器产业株式会社 | Life span estimation method for a lead acid battery and a power supply system |
| JP2010117256A (en) * | 2008-11-13 | 2010-05-27 | Toshiba Toko Meter Systems Co Ltd | Battery lifetime prediction apparatus |
| CN103323690A (en) * | 2012-03-23 | 2013-09-25 | 海洋王（东莞）照明科技有限公司 | Service life testing method of lamp charger and service life testing device |
| CN102778653A (en) * | 2012-06-20 | 2012-11-14 | 哈尔滨工业大学 | Data-driven lithium ion battery cycle life prediction method based on AR (Autoregressive) model and RPF (Regularized Particle Filtering) algorithm |
| CN103344923A (en) * | 2013-08-01 | 2013-10-09 | 哈尔滨工业大学 | EKF (extended Kalmar filter)-method and NSDP-AR-model-based method for predicting cycling life of fusion-type lithium ion battery |
| CN103399279A (en) * | 2013-08-01 | 2013-11-20 | 哈尔滨工业大学 | Method for predicting cycle life of fused lithium ion battery based on EKF (Extended Kalman Filter) method and AR (AutoRegressive) model |
| CN103399281A (en) * | 2013-08-01 | 2013-11-20 | 哈尔滨工业大学 | Lithium ion battery cycle life predicating method based on cycle life degeneration stage parameter ND-AR (neutral density-autoregressive) model and EKF (extended Kalman filter) method |
| CN103954913A (en) * | 2014-05-05 | 2014-07-30 | 哈尔滨工业大学深圳研究生院 | Predication method of electric vehicle power battery service life |
| CN104714189A (en) * | 2015-04-02 | 2015-06-17 | 奇瑞汽车股份有限公司 | Method for predicting cycle life of battery pack for electric car |
Non-Patent Citations (3)
| Title |
|---|
| 彭喜元 等: "《数据驱动的故障预测》", 31 March 2016, 哈尔滨工业大学出版社 * |
| 彭宇 等: "数据驱动故障预测和健康管理综述", 《仪器仪表学报》 * |
| 艾力 等: "基于数据驱动的卫星锂离子电池寿命预测方法", 《计算机测量与控制》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110058177A (en) * | 2019-05-06 | 2019-07-26 | 奇瑞新能源汽车技术有限公司 | A kind of power battery electricity SOC modification method |
| CN110058177B (en) * | 2019-05-06 | 2021-08-31 | 奇瑞新能源汽车股份有限公司 | Power battery electric quantity SOC correction method |
| CN110658462A (en) * | 2019-10-25 | 2020-01-07 | 北京航空航天大学 | Lithium battery online service life prediction method based on data fusion and ARIMA model |
| CN110658462B (en) * | 2019-10-25 | 2020-12-15 | 北京航空航天大学 | Lithium battery online service life prediction method based on data fusion and ARIMA model |
| CN110888077A (en) * | 2019-10-30 | 2020-03-17 | 无锡市产品质量监督检验院 | Accelerated lithium ion battery life evaluation method based on ARIMA time sequence |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107544033B (en) | 2020-01-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hua et al. | Toward sustainable reuse of retired lithium-ion batteries from electric vehicles | |
| Xu et al. | Remaining useful life prediction of lithium-ion batteries based on wiener process under time-varying temperature condition | |
| Gou et al. | State-of-health estimation and remaining-useful-life prediction for lithium-ion battery using a hybrid data-driven method | |
| Xiong et al. | An electrochemical model based degradation state identification method of Lithium-ion battery for all-climate electric vehicles application | |
| Li et al. | A practical lithium-ion battery model for state of energy and voltage responses prediction incorporating temperature and ageing effects | |
| Khan et al. | Batteries state of health estimation via efficient neural networks with multiple channel charging profiles | |
| Ghorbanzadeh et al. | Long-term degradation based analysis for lithium-ion batteries in off-grid wind-battery renewable energy systems | |
| Yayan et al. | A novel method for SoH prediction of batteries based on stacked LSTM with quick charge data | |
| Pei et al. | An Equivalent Circuit Model for Lithium Battery of Electric Vehicle considering Self‐Healing Characteristic | |
| CN107544033A (en) | Digital-analog fusion prediction method for remaining service life of lithium ion battery | |
| Wang et al. | Health diagnosis for lithium-ion battery by combining partial incremental capacity and deep belief network during insufficient discharge profile | |
| Sun et al. | Adaptive evolution enhanced physics-informed neural networks for time-variant health prognosis of lithium-ion batteries | |
| Jafari et al. | Prediction of the battery state using the digital twin framework based on the battery management system | |
| Shao et al. | A novel method of discharge capacity prediction based on simplified electrochemical model-aging mechanism for lithium-ion batteries | |
| Liang et al. | Evaluation of battery modules state for electric vehicle using artificial neural network and experimental validation | |
| Semeraro et al. | Digital twin in battery energy storage systems: Trends and gaps detection through association rule mining | |
| De Sousa et al. | Comparison of different approaches for lead acid battery state of health estimation based on artificial neural networks algorithms | |
| Garg et al. | Reconfigurable battery systems: Challenges and safety solutions using intelligent system framework based on digital twins | |
| Dini et al. | Review on modeling and soc/soh estimation of batteries for automotive applications | |
| Swarnkar et al. | Electric vehicle lithium-ion battery ageing analysis under dynamic condition: A machine learning approach | |
| Pande et al. | Electrical models for EV’s batteries: an overview and mathematical design of RC network | |
| Huang et al. | A multi-variable multi-step Seq2seq networks for the state of charge estimation of lithium-ion battery | |
| Catelani et al. | Remaining Useful Life estimation for electric vehicle batteries using a similarity-based approach | |
| Chu et al. | Parameter identification and SOC estimation of lithium-ion batteries based on AGCOA-ASRCKF | |
| Dineva et al. | Data-driven terminal voltage prediction of li-ion batteries under dynamic loads |
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 |