CN103389471B - A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section - Google Patents
A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section Download PDFInfo
- Publication number
- CN103389471B CN103389471B CN201310317281.9A CN201310317281A CN103389471B CN 103389471 B CN103389471 B CN 103389471B CN 201310317281 A CN201310317281 A CN 201310317281A CN 103389471 B CN103389471 B CN 103389471B
- Authority
- CN
- China
- Prior art keywords
- battery
- pressure drop
- gpr
- capacity
- discharge time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
Based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section, the present invention relates to a kind of battery life predicting method.The invention solves the problem that existing method cannot realize lithium battery cycle life prediction, the present invention adopts ESN algorithm, carry out degeneration modeling, adopt the modeling method that Gaussian process returns, the degradation model training waiting pressure drop forecast model discharge time to carry out based on ESN of setting up based on GPR is trained with pressure drop forecast model discharge time that waits based on GPR, acquisition waits pressure drop forecast model discharge time, carry out forecast model discharge time such as pressure drop such as grade based on GPR, acquisition waits the predicted value of pressure drop discharge time; Carry out the degradation model based on ESN, obtain lower N
1the discharge capacity of the battery of individual discharge cycle; The remaining capacity value of battery compares with the failure threshold row of battery capacity, completes the indirect predictions of battery cycle life.The present invention is applicable to battery life predicting.
Description
Technical field
The present invention relates to a kind of battery life predicting method.
Background technology
Although lithium ion battery is a kind of stored energy and conversion equipment, it is not infinitely to use, and namely its service life cycle is limited, this is because the performance of battery can decline gradually along with the use of battery.
Lithium ion battery is a kind of rechargeable battery, it mainly rely on lithium ion between a positive electrode and a negative electrode movement carry out work, the chemomotive force of whole battery comes from the difference of its two electrode chemical potential.Convert electrical energy into chemical energy during charge in batteries to store in the battery, during electric discharge, then chemical energy is converted to electric energy for load.Due to the reversibility of two kinds of energy conversion, it is unlimited for seeming the cyclic process of discharge and recharge, actually this is not so, this is because in the cyclic process of discharge and recharge, some irreversible processes can be there are in inside battery, cause the change of internal driving, output current etc., cause the decay of battery capacity, thus have impact on the service life cycle of battery.
Lithium ion battery is in cycle charge discharge electric process, and some irreversible chemical reaction processes can occur inside battery, cause the loss of the Li+ that electrode " embeds/deviates from ", thus internal battery impedance is improved, and directly translate into the decline of battery open circuit voltage.
Utilize resistive impedance spectrometry to record internal resistance of cell impedance and comprise charge transfer resistance RCT, Warburg impedance RW and bath resistance RE, wherein the impact of Warburg impedance RW on cell degradation process is insignificant, therefore can ignore.The experimental data that the PCoE research centre of NASA is a large amount of by analysis finds, there is between battery capacity and internal driving the linear dependence of height, battery capacity will be degenerated gradually along with the ageing process of battery, namely the battery capacity after each charge and discharge cycles can decline gradually, thus do not reach rated capacity, therefore the degeneration of battery capacity can be utilized as the main in circulating battery serviceable life, but the historical data had due to the prediction of service life of lithium battery is few, the difficult foundation of model, probabilistic shortcoming, and cannot realize lithium battery cycle life prediction.
Summary of the invention
The present invention cannot realize the problem of lithium battery cycle life prediction in order to solve existing method, propose a kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section.
Of the present invention a kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section, the concrete steps of the method are:
The electricity z that step one, the collection charging-discharging cycle number of times x of mesuring battary, the sparking voltage of each charging-discharging cycle and battery capacity and each charging-discharging cycle are released,
Step 2, according to gathering the charging-discharging cycle number of times x of mesuring battary and the sparking voltage of each charging-discharging cycle and battery capacity, to calculate the pressure drop such as corresponding poor for discharge time, sequences y discharge time such as pressure drop such as acquisition grade;
Step 3, adopt ESN algorithm, the residual capacity data z of the battery after pressure drop sequences y discharge time such as utilization and each discharge and recharge carries out degeneration modeling, obtains the degradation model based on ESN;
Step 4, adopt Gaussian process return modeling method, utilize charging-discharging cycle number of times x and the battery charging and discharging cycle corresponding wait pressure drop sequences y discharge time foundation based on GPR etc. pressure drop forecast model discharge time;
Step 5, by etc. pressure drop sequence data discharge time y and each discharge cycle data set of electricity z of releasing
carry out training based on the degradation model of ESN as training set, by the charging-discharging cycle number of times x of battery and the data set waiting pressure drop sequences y discharge time
carry out training based on pressure drop forecast model discharge time that waits of GPR as training data, acquisition waits pressure drop forecast model discharge time, and wherein N is positive integer;
Step 6, by lower N
1individual charging-discharging cycle time manifold
input forecast model discharge time such as pressure drop such as grade based on GPR, acquisition waits the predicted value of pressure drop discharge time
Step 7, by the predicted value of the pressure drop such as acquisition discharge time
substitute into the degradation model based on ESN, obtain lower N
1the discharge capacity of the battery of individual discharge cycle
Step 8, the initial capacity of battery is deducted lower N
1the discharge capacity of the battery of individual charging-discharging cycle
after the remaining capacity value of battery and the failure threshold of battery capacity compare, judge whether the remaining capacity value of battery equals the failure threshold of battery capacity, then using the residual life of charging-discharging cycle N as battery, complete based on the indirect predictions of GPR with the cycle life of lithium ion battery of indeterminacy section, otherwise perform step 9;
Step 9, the remaining capacity value of battery and the failure threshold of battery capacity to be compared, if the remaining capacity value of battery is greater than the failure threshold of battery capacity, then make N=N+N
1, return execution step 5, if the remaining capacity value of battery is less than the failure threshold of battery capacity, then make N=N-N
2, return execution step 5, wherein N
2for being less than N
1positive integer.
The modeling method that the present invention adopts ESN algorithm and Gaussian process to return is combined, pressure drop sequential forecasting models discharge time such as GPR algorithm foundation are adopted to predict sequence discharge time such as pressure drop such as grade of future time instance, what finally prediction obtained waits pressure drop sequence inputting discharge time to the degradation model of lithium ion battery, thus realizes lower N
1the prediction of the battery capacity in individual moment, and then realize cycle life of lithium ion battery indirect predictions.
Accompanying drawing explanation
Fig. 1 is based on the degeneration modeling checking curve map of the NASA lithium ion battery of ESN, and in figure, curve 1 is estimated value curve, and curve 2 is actual value curve;
Fig. 2 is the capacity of lithium ion battery estimated value that modeling obtains and the error curve diagram recorded between capacity actual value;
Fig. 3 is the prediction effect figure that employing 30% data carry out model training, 3 real remaining lifes when being 80% failure threshold in figure, 4 is fiducial interval, 5 prediction averages when being 80% failure threshold, 6 is 80% failure threshold, and 7 is fiducial interval, 8 prediction averages when being 70% failure threshold, 9 is 70% failure threshold, 10 real remaining lifes when being 80% failure threshold;
Fig. 4 is the prediction effect figure that employing 50% data carry out model training;
Fig. 5 adopts 70% data to carry out the prediction effect figure of model training.
Embodiment
A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section described in embodiment one, present embodiment, the concrete steps of the method are:
The electricity z that step one, the collection charging-discharging cycle number of times x of mesuring battary, the sparking voltage of each charging-discharging cycle and battery capacity and each charging-discharging cycle are released,
Step 2, according to gathering the charging-discharging cycle number of times x of mesuring battary and the sparking voltage of each charging-discharging cycle and battery capacity, to calculate the pressure drop such as corresponding poor for discharge time, sequences y discharge time such as pressure drop such as acquisition grade;
Step 3, adopt ESN algorithm, the residual capacity data z of the battery after pressure drop sequences y discharge time such as utilization and each discharge and recharge carries out degeneration modeling, obtains the degradation model based on ESN;
Step 4, adopt Gaussian process return modeling method, utilize charging-discharging cycle number of times x and the battery charging and discharging cycle corresponding wait pressure drop sequences y discharge time foundation based on GPR etc. pressure drop forecast model discharge time;
Step 5, by etc. pressure drop sequence data discharge time y and each discharge cycle data set of electricity z of releasing
carry out training based on the degradation model of ESN as training set, by the charging-discharging cycle number of times x of battery and the data set waiting pressure drop sequences y discharge time
carry out training based on pressure drop forecast model discharge time that waits of GPR as training data, acquisition waits pressure drop forecast model discharge time, and wherein N is positive integer;
Step 6, by lower N
1individual charging-discharging cycle time manifold
input forecast model discharge time such as pressure drop such as grade based on GPR, acquisition waits the predicted value of pressure drop discharge time
Step 7, by the predicted value of the pressure drop such as acquisition discharge time
substitute into the degradation model based on ESN, obtain lower N
1the discharge capacity of the battery of individual discharge cycle
Step 8, the initial capacity of battery is deducted lower N
1the discharge capacity of the battery of individual charging-discharging cycle
after the remaining capacity value of battery and the failure threshold of battery capacity compare, judging whether the remaining capacity value of battery equals the failure threshold of battery capacity, is then by charging-discharging cycle N
1as the residual life of battery, complete based on the indirect predictions of GPR with the cycle life of lithium ion battery of indeterminacy section, otherwise perform step 9;
Step 9, the remaining capacity value of battery and the failure threshold of battery capacity to be compared, if the remaining capacity value of battery is greater than the failure threshold of battery capacity, then make N=N+N
1, return execution step 5, if the remaining capacity value of battery is less than the failure threshold of battery capacity, then make N=N-N
2, return execution step 5, wherein N
2for being less than N
1positive integer.
Present embodiment adopts pressure drop sequential forecasting models discharge time such as GPR algorithm foundation to predict sequence discharge time such as pressure drop such as grade of future time instance, what finally prediction obtained waits pressure drop sequence inputting discharge time to the degradation model of lithium ion battery, thus the prediction realized future time instance capacity.
Embodiment two, present embodiment are a kind of based on the further illustrating of cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section to described in embodiment one, adopt ESN algorithm described in step 3, the method that the residual capacity data z of pressure drop sequences y discharge time such as utilization and battery carries out degeneration modeling is:
The method of step 3 one, employing cross validation, utilizes and obtains deposit pond scale N, spectral radius sr, input block yardstick and input block displacement, and obtain the output weights of ESN;
Step 3 two, use with Monotone constraint quadratic programming equation training ESN output weights, make battery capacity estimation value
and the error sum of squares between actual value y (n) is minimum, complete degeneration modeling.
ESN is a kind of method of black box, and its modeling result does not provide with expression form.Its modeling process comprises two parts: one is ESN training process, be about to part and input data: wait pressure drop sequence x discharge time (n), export data: battery capacity y (n) is as training set, carry out ESN training, thus obtain the degradation model based on ESN.In ESN model, having 4 parameter influence performance of modelings, is deposit pond scale N, spectral radius sr, input block yardstick (InputScaling, IS) and input block displacement (InputShift, IF) respectively.Training process is exactly the optimal value adopting the method for cross validation to obtain above-mentioned 4 parameters, and uses the output weights of the quadratic programming equation training ESN with Monotone constraint, thus makes battery capacity estimation value
and the error sum of squares between actual value y (n) is minimum.Two is modelling verification processes, brings degradation model into, calculates battery capacity estimation value, and this estimated value and True Data are analyzed by remaining input data, thus the accuracy of checking degradation model.
In sum, cell degradation modeling process based on ESN is exactly the process of 4 the parameter optimal values determining ESN according to training data, namely pond scale N, spectral radius sr, input block yardstick (InputScaling is laid in, and input block displacement (InputShift, IF) IS).These four parameters are once determine, degradation model also just determines, and just its concrete expression formula cannot provide.
Model error described in present embodiment: use root-mean-square error (RootMeanSquaredError, RMSE) as the evaluation index of approximation capability, as formula:
In formula, n is the length of training data or test data,
for the output valve of ESN, i.e. battery remaining power predicted value, y (i) is i-th battery remaining power actual value
Overall fit effect: adopt R
2the overall fit effect of evaluation function, when the fitting effect of model non-constant time, the quadratic sum of the error of model output valve and actual value can be greater than the error sum of squares of the average of model output valve and actual value, i.e. R
2negative value may be there is; As formula, in formula,
for
average.
Degeneration modeling result:
Shown in the degradation model parameter obtained according to above-mentioned modeling process and training process degradation model evaluation result table 1.
Table 1 degradation model result and model-evaluation index
Degradation model is verified:
Obtain 4 parameters of the lithium ion battery degradation model based on ESN as shown above, from the degradation model obtaining lithium ion battery, and by calculating the evaluation index of training process model.The accuracy of degeneration modeling is verified, pressure drop sequence discharge time will be waited
bring cell degradation model into, by the capability value of model assessment battery, and this estimated value and actual value are analyzed, thus the accuracy of verification model, its checking is as shown in Figure 1.
In Fig. 1, curve 1 is the capacity of lithium ion battery estimated value curve calculated based on ESN degradation model, and what dotted line 2 represented is the actual value curve of battery capacity.Fig. 2 is the capacity of lithium ion battery estimated value that modeling obtains and the error curve diagram recorded between capacity actual value.
Root-mean-square error between calculated capacity estimated value and actual value and R
2result is as shown in table 2.
Table 2 is based on the degradation model checking evaluation index of ESN
In sum, pressure drop sequences discharge time such as employing proposed by the invention can the capacity of characterizing battery, and by the degeneration modeling achieving battery of ESN algorithm, Fig. 1 demonstrates the accuracy of cell degradation model, and model error is between-0.04 ~ 0.12 as can be seen from Figure 2.The overall fit effect of the root-mean-square error provided from table 2 and model also can show the validity of degenerate state modeling method in this paper.
Embodiment three, present embodiment are a kind of based on the further illustrating of cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section to described in embodiment one, the modeling method that employing Gaussian process described in step 4 returns, utilize charging-discharging cycle number of times x and the battery charging and discharging cycle corresponding etc. pressure drop y discharge time sequence, the method for pressure drop forecast model discharge time that waits obtained based on GPR is:
Step 4 one, the charging-discharging cycle number of times x extracting mesuring battary, the sparking voltage of each charging-discharging cycle and the partitioned data set (PDS) of battery capacity
carry out forecast model training, wherein N is positive integer;
Step 4 two, training data is inputted GPR model, carry out the training of GPR forecast model, obtain GPR forecast model;
Step 4 three, basis obtain GPR forecast model, by the charging-discharging cycle of future time instance
input prediction model, obtains the capacity predict value in lower N number of moment
and variance, obtain forecast model discharge time such as pressure drop such as grade based on GPR.
Deng pressure drop sequence prediction discharge time be utilize known charging-discharging cycle number of times and charging-discharging cycle corresponding etc. pressure drop sequence data discharge time carry out forecast model training, obtain an optimum forecast model, wait pressure drop discharge time then with this model extrapolation following some cycles.Because data inevitably introduce noise in gatherer process, make data have uncertainty, take into full account this point, the present invention adopts Gaussian process recurrence (GPR) algorithm to carry out data test and prognostic experiment.Gaussian process regression model (GPR) be one flexibly, there is uncertain nonparametric model of expressing, and, GPR can carry out modeling by the combination of the kernel function of suitable Gaussian process to the behavior of arbitrary system, final realization based on the prediction of Bayesian forecasting framework, in this process can be flexible in conjunction with priori.Gaussian process predict the outcome while prediction of output result, the variance of prediction can also be provided, namely determine prediction confidence intervals, add the accuracy of prediction.Now, it has become a very important part in battery status prediction and health control algorithm.
The input data of GPR model are charging-discharging cycle number of times, discharge time such as pressure drop such as input data etc.GPR forecast model has two committed steps, and one is model training, and two is model predictions, introduces respectively below.
GPR model training
The thought of Gaussian process modeling is exactly the parametrization or the imparametrization form that do not need to provide f (x) in y=f (x), directly in function space, regard the value of f (x) as stochastic variable, regard the prior probability distribution p (f (x)) of f (x) as Gaussian distribution.If data-oriented collection
and define input data matrix X ∈ R
d × N, export data vector y ∈ R
n × 1.In the limited data set of data-oriented collection D, f (x
1) ..., f (x
n) set (stochastic variable is regarded in each set as) of stochastic variable can be formed, and there is Joint Gaussian distribution, the stochastic process that their are formed just is referred to as Gaussian process.Namely
f(x)～GP(m(x),k(x
i,x
j))(3)
Wherein, m (x)=E [f (x)], k (x
i, x
j)=E [(f (x
i)-m (x
i) (f (x
j)-m (x
j))], symbol E represents mathematical expectation.M (x) is mean value function, k (x
i, x
j) be covariance function.
Gaussian process is applied to general regression modeling problem, the observed object value y of Noise can be considered, namely
y＝f(x)+ε(4)
Wherein ε is additional independently white Gaussian noise incoherent with f (x), namely obeys that average is zero, variance is
normal distribution, can be denoted as
for (4) formula, because noise ε is the white Gaussian noise independent of f (x), if f (x) is Gaussian process, then the same Gaussian distributed of y, the set of its finite observation value joint distribution can form a Gaussian process, namely
Wherein, m (x) is mean value function, δ
ijdirac function, during i=j, function δ
ij=1; I and j is respectively i-th and a jth input variable.
Y=f (x)+ε, it is the function expression for predicting, different from general regression problem, f (x) can not show by parameter or non-parametric form, and known be exactly f (x) be a Gaussian process, wherein each variable f (x
1) ..., f (x
n) obey Joint Gaussian distribution, so each training points is brought into by forecast model exactly that obtain obtain matrix
so the form that forecast model is write as matrix is as follows:
In 6 formulas, I represents the unit matrix of N × N, and C (X, X) represents the covariance matrix of N × N, and K (X, X) represents the nuclear matrix of N × N, is called Gram matrix, its element k
ij=k (x (i), x (j)).
Here (6) formula can be understood as the relation (being equivalent to the y=ax+b in one-variable linear regression) between y and x.Wherein m (x) with
all contain unknown parameter, be referred to as hyper parameter, as m (x)=a+bx,
Hyper parameter is Θ=[a, b, υ
0, ω
l, σ
n] a in these hyper parameter phase one-variable linear regressions, b is that unknown number is determined by training data, υ
0for the variance of covariance function, ω
lfor the distance size of covariance function, σ
nfor the variance of noise.
GPR model prediction
In the functional space that Gaussian process prior distribution defines, the function prediction output valve of Posterior distrbutionp can be calculated under Bayesian frame.When predicting, for N
*the set of individual input data (charging-discharging cycle)
input data matrix X is formed by input
*∈ R
d × N*, corresponding prediction output has average
and variance
gaussian distribution, namely
Confidence level or the uncertainty of model prediction can also be provided while providing prediction of output value, wherein from formula 7, GPR model
for the covariance function matrix of test data and training data,
for the covariance function matrix of training data, y is the object vector of training data, k
*=k (x
*, x
*) be the covariance function of test data.
Embodiment four, present embodiment are a kind of based on the further illustrating of cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section to described in embodiment three, and the region that the variance described in step 4 three covers is the fiducial interval of 95%.
Checking and analysis
The checking of IPC10 (cycle2) battery and analysis
Adopt 30% of total data, 50% and 70% to carry out forecast model training respectively, remaining data is analyzed with predicted value as checking collection.Due to, ICIP10 battery has only carried out the volume test in every 500 cycles, so present embodiment only carries out the comparative analysis in corresponding moment;
Experimentation is as follows:
Data set: gather IIPC10 (cycle2) battery data and corresponding charging-discharging cycle number, build data set
x is charging-discharging cycle, and y is discharge time such as pressure drop such as grade, and z is battery capacity;
Model training: will
carry out training based on the degradation model of ESN as training set, then will
carry out training based on pressure drop forecast model discharge time that waits of GPR as training data, obtain waiting pressure drop forecast model discharge time.
Prediction: by lower N
*the charged electrical cycle data collection in individual moment
the pressure drop such as input forecast model discharge time carries out the predicted value waiting pressure drop discharge time
then this predicted value is substituted into cell degradation model and obtain battery capacity prediction value
Model analysis: by capacity predict value
with actual value
be analyzed;
Experimental result:
In order to verify validity and the adaptability of put forward the methods herein, the data of total data 30%, 50% and 70% are adopted to carry out modeling as training data, capacity predict value, fiducial interval, the predicated error of result corresponding to each cycle period respectively.
Experimental result as shown in Table 3-5.
The battery capacity prediction of table 330% training data
The battery capacity prediction of table 450% training data
The battery capacity prediction of table 570% training data
From table 3-5, what the present invention proposed a kind ofly achieves prediction to battery capacity based on GPR with the cycle life of lithium ion battery indirect predictions method of indeterminacy section, while prediction, give the fiducial interval predicted the outcome, and demonstrate the adaptability of the method by different training set.
Embodiment five, present embodiment are a kind of based on the further illustrating of cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section to described in embodiment one, and in step 8, the failure threshold of battery capacity is 70% or 80% of the initial capacity of battery.
The RUL Forecasting Methodology of battery is consistent with capacity prediction methods, and the result just exported is different.Adopt 70% and 80% of initial capacity respectively as failure threshold.Adopt 30% of total data, 50% and 70% to carry out modeling during forecast model modeling respectively, provide the RUL predicted value of battery, fiducial interval, error.Adopt 30% data to carry out the prediction effect of model training as shown in Figure 3, adopt 50% data to carry out the prediction effect of model training as shown in Figure 4.Adopt 70% data to carry out the prediction effect of model training as shown in Figure 5, the failure threshold due to 80% has been included in training data and has suffered, so can only carry out the RUL prediction of 70% failure prediction.The RUL of three kinds of training data length predicts the outcome as shown in table 6.
The RUL of table 6 three kinds of training data length predicts the outcome
The checking of NASA battery and analysis
No. 18 batteries of the BatteryDataSet experimental data adopting NASA to provide carry out experimental verification.
This data set derives from the lithium ion battery test envelope built in NASAPCoE research centre, battery charging, electric discharge and impedance measurement experiment, runs at room temperature 25 DEG C:
Be charge, until cell voltage reaches 4.2V under the pattern of 1.5A at steady current;
Be discharge, until cell voltage drops to 2.5V under the pattern of 2A at steady current;
Measure battery impedance by EIS, the scope of frequency sweeping is from 0.1Hz to 5kHz.
No. 18 battery data collection totally 132 capacity datas, in order to verify the validity of the Forecasting Methodology that the present invention proposes, adopt the training set of 30%, 50% and 70% 3 kind of length of all told data to carry out the training of forecast model respectively, remaining data is used for checking and the comparative analysis of model.
One, battery capacity prediction
Adopt the method for the invention to predict battery capacity, predict the outcome as shown in table 7-table 9.
The capacity predict result of table 730% training data
The capacity predict result of table 850% training data
The capacity predict result of table 970% training data
Two, remaining battery life prediction
The Forecasting Methodology of the residual life of NANSA battery is adopt 70% and 80% of initial capacity respectively as failure threshold, 30% of total data, 50% and 70% is adopted to carry out modeling during forecast model modeling respectively, provide the RUL predicted value of battery, fiducial interval, error.The RUL of three kinds of training data length predicts the outcome as shown in table 10.
The RUL of table 10 three kinds of training data length predicts the outcome
The lithium ion battery that the method for the invention solves application on site cannot realize capacity predict and predicting residual useful life problem; The method can not only provide the point estimate predicted the outcome, and gives the fiducial interval predicted the outcome, and makes to predict the outcome more reasonable, larger to the directive significance of user.
Claims (5)
1., based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section, it is characterized in that, the concrete steps of the method are:
The electricity z that step one, the collection charging-discharging cycle number of times x of mesuring battary, the sparking voltage of each charging-discharging cycle and battery capacity and each charging-discharging cycle are released,
Step 2, according to gathering the charging-discharging cycle number of times x of mesuring battary and the sparking voltage of each charging-discharging cycle and battery capacity, to calculate the pressure drop such as corresponding poor for discharge time, y such as sequence data discharge time such as pressure drop such as acquisition grade;
Step 3, adopt ESN algorithm, the residual capacity data z of the battery after pressure drop sequences y discharge time such as utilization and each discharge and recharge carries out degeneration modeling, obtains the degradation model based on ESN;
Step 4, adopt Gaussian process return modeling method, utilize charging-discharging cycle number of times x and the battery charging and discharging cycle corresponding wait pressure drop sequences y discharge time foundation based on GPR etc. pressure drop forecast model discharge time;
Step 5, by etc. pressure drop sequence data discharge time y and each discharge cycle data set of electricity z of releasing
carry out training based on the degradation model of ESN as training set, by the charging-discharging cycle number of times x of battery and the data set waiting pressure drop sequences y discharge time
carry out training based on pressure drop forecast model discharge time that waits of GPR as training data, acquisition waits pressure drop forecast model discharge time, and wherein N is positive integer;
Step 6, by lower N
1individual charging-discharging cycle time manifold
input forecast model discharge time such as pressure drop such as grade based on GPR, acquisition waits the predicted value of pressure drop discharge time
Step 7, by the predicted value of the pressure drop such as acquisition discharge time
substitute into the degradation model based on ESN, obtain lower N
1the discharge capacity of the battery of individual discharge cycle
Step 8, the initial capacity of battery is deducted lower N
1the discharge capacity of the battery of individual charging-discharging cycle
after the remaining capacity value of battery and the failure threshold of battery capacity compare, judge whether the remaining capacity value of battery equals the failure threshold of battery capacity, then using the residual life of charging-discharging cycle N as battery, complete based on the indirect predictions of GPR with the cycle life of lithium ion battery of indeterminacy section, otherwise perform step 9;
Step 9, the remaining capacity value of battery and the failure threshold of battery capacity to be compared, if the remaining capacity value of battery is greater than the failure threshold of battery capacity, then make N=N+N
1, return execution step 5, if the remaining capacity value of battery is less than the failure threshold of battery capacity, then make N=N-N
2, return execution step 5, wherein N
2for being less than N
1positive integer.
2. according to claim 1 a kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section, it is characterized in that, adopt ESN algorithm described in step 3, the method that the residual capacity data z of the battery after the pressure drop sequence data discharge time y such as utilization and each discharge and recharge carries out degeneration modeling is:
The method of step 3 one, employing cross validation, utilizes and obtains deposit pond scale N, spectral radius sr, input block yardstick and input block displacement, and obtain the output weights of ESN;
Step 3 two, use with Monotone constraint quadratic programming equation training ESN output weights, make battery capacity estimation value
and the error sum of squares between actual value y (n) is minimum, complete degeneration modeling.
3. according to claim 1 a kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section, it is characterized in that, the modeling method that employing Gaussian process described in step 4 returns, utilize charging-discharging cycle number of times x and the battery charging and discharging cycle corresponding etc. the data of pressure drop y discharge time sequence, the method for pressure drop forecast model discharge time that waits obtained based on GPR is:
Step 4 one, the charging-discharging cycle number of times x extracting mesuring battary, the sparking voltage of each charging-discharging cycle and the partitioned data set (PDS) of battery capacity
carry out forecast model training, wherein N is positive integer;
Step 4 two, training data is inputted GPR model, carry out the training of GPR forecast model, obtain GPR forecast model;
Step 4 three, basis obtain GPR forecast model, by the charging-discharging cycle of future time instance
input prediction model, obtains the capacity predict value in lower N number of moment
and variance, obtain forecast model discharge time such as pressure drop such as grade based on GPR.
4. according to claim 3 a kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section, it is characterized in that, the region that the variance described in step 4 three covers is the fiducial interval of 95%.
5. according to claim 1ly a kind ofly to it is characterized in that based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section, in step 8, the failure threshold of battery capacity is 70% or 80% of the initial capacity of battery.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310317281.9A CN103389471B (en) | 2013-07-25 | 2013-07-25 | A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310317281.9A CN103389471B (en) | 2013-07-25 | 2013-07-25 | A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103389471A CN103389471A (en) | 2013-11-13 |
| CN103389471B true CN103389471B (en) | 2015-12-09 |
Family
ID=49533798
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310317281.9A Active CN103389471B (en) | 2013-07-25 | 2013-07-25 | A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103389471B (en) |
Families Citing this family (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103954915B (en) * | 2014-05-16 | 2017-02-01 | 哈尔滨工业大学 | Lithium ion battery remaining life indirect prediction method based on probability integration |
| CN106339755B (en) * | 2016-08-29 | 2018-09-21 | 深圳市计量质量检测研究院 | Lithium battery health status prediction technique based on neural network Yu period kernel function GPR |
| CN106405427A (en) * | 2016-08-29 | 2017-02-15 | 哈尔滨理工大学 | Lithium battery state of health prediction method based on neural network and Maternard kernel function GPR |
| CN106125004A (en) * | 2016-08-29 | 2016-11-16 | 哈尔滨理工大学 | Lithium battery health status Forecasting Methodology based on neutral net kernel function GPR |
| CN106777786A (en) * | 2017-01-12 | 2017-05-31 | 常州工学院 | A kind of lithium ion battery SOC estimation method |
| CN107015158B (en) * | 2017-04-01 | 2019-05-24 | 湖南银杏数据科技有限公司 | Lithium battery remaining life on-line quick detection method based on constant-current discharge segment |
| CN108037463B (en) * | 2017-12-15 | 2020-01-21 | 太原理工大学 | Lithium ion battery life prediction method |
| CN107942261B (en) * | 2017-12-29 | 2021-04-02 | 上海电气集团股份有限公司 | Method and system for estimating state of charge of battery |
| CN108680861B (en) * | 2018-03-05 | 2021-01-01 | 北京航空航天大学 | Method and device for constructing lithium battery residual cycle life prediction model |
| KR102291133B1 (en) * | 2018-03-07 | 2021-08-20 | 주식회사 엘지화학 | Apparatus and method for predict battery life |
| CN109271670A (en) * | 2018-08-16 | 2019-01-25 | 天津大学 | CMOS low-noise amplifier performance degradation prediction technique based on BPNN |
| CN109101735A (en) * | 2018-08-16 | 2018-12-28 | 天津大学 | A kind of cmos circuit performance degradation prediction technique based on RNN |
| CN109284543B (en) * | 2018-09-04 | 2023-05-23 | 河北工业大学 | IGBT residual life prediction method based on optimal scale Gaussian process model |
| CN109298351B (en) * | 2018-09-30 | 2021-07-27 | 清华大学深圳研究生院 | New energy vehicle-mounted battery residual life estimation method based on model learning |
| CN109031153B (en) * | 2018-10-16 | 2020-01-24 | 北京交通大学 | Online health state estimation method for lithium ion battery |
| JP2020137156A (en) * | 2019-02-13 | 2020-08-31 | トヨタ自動車株式会社 | Vehicle, information terminal and vehicle control method |
| CN109782192A (en) * | 2019-03-08 | 2019-05-21 | 安徽理工大学 | Lithium ion battery residual life prediction technique under different discharge-rates |
| CN110161425B (en) * | 2019-05-20 | 2020-05-19 | 华中科技大学 | Method for predicting remaining service life based on lithium battery degradation stage division |
| CN110222431B (en) * | 2019-06-11 | 2022-04-12 | 哈尔滨工业大学 | Lithium ion battery residual life prediction method based on gate control cycle unit neural network and Kalman filtering model fusion |
| CN112084459A (en) * | 2019-06-12 | 2020-12-15 | 电计贸易(上海)有限公司 | Method and device for predicting battery charge-discharge cycle life, electronic terminal and storage medium |
| CN111143973B (en) * | 2019-12-05 | 2021-01-26 | 云南电网有限责任公司玉溪供电局 | Valve-regulated lead-acid storage battery degradation trend prediction method based on Gauss process regression |
| CN111460380B (en) * | 2020-03-30 | 2022-03-18 | 上海交通大学 | Multi-working-condition driving range prediction method and system based on Gaussian process regression |
| WO2021208079A1 (en) * | 2020-04-17 | 2021-10-21 | 中国科学院深圳先进技术研究院 | Method and apparatus for obtaining power battery life data, computer device, and medium |
| CN111832221B (en) * | 2020-06-23 | 2023-09-26 | 华中科技大学无锡研究院 | Lithium battery life prediction method based on feature screening |
| CN112098848B (en) * | 2020-09-14 | 2023-04-11 | 北京大学深圳研究生院 | Battery discharge capacity prediction method and system and readable storage medium |
| CN112255559B (en) * | 2020-10-12 | 2023-05-09 | 启东沃太新能源有限公司 | Lithium battery energy storage power station residual life prediction method based on multiple linear regression |
| CN112798967B (en) * | 2020-12-04 | 2022-03-15 | 电子科技大学 | Long-term and short-term prediction method for solid oxide fuel cell |
| CN113820608B (en) * | 2021-08-20 | 2022-08-05 | 北京邮电大学 | Method for predicting remaining capacity of battery in echelon and electronic equipment |
| CN113900033B (en) * | 2021-09-23 | 2022-07-08 | 浙江大学 | Lithium battery online service life prediction method based on charging data spatial distribution characteristics |
| CN114167284B (en) * | 2021-11-02 | 2023-12-22 | 江苏博强新能源科技股份有限公司 | Lithium battery RUL prediction method and equipment based on BMS big data and integrated learning |
| CN117374437B (en) * | 2023-12-07 | 2024-03-19 | 天津国能津能滨海热电有限公司 | Storage battery life management system, method, device, storage medium and equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1813199A (en) * | 2003-06-25 | 2006-08-02 | 罗伯特.博世有限公司 | Method for predicting the residual service life of an electric energy accumulator |
| CN102262211A (en) * | 2011-04-20 | 2011-11-30 | 哈尔滨工业大学 | Analog circuit failure diagnosis method based on dynamic classification of echo state network |
| CN102798823A (en) * | 2012-06-15 | 2012-11-28 | 哈尔滨工业大学 | Gaussian process regression-based method for predicting state of health (SOH) of lithium batteries |
| CN103033761A (en) * | 2012-12-17 | 2013-04-10 | 哈尔滨工业大学 | Lithium ion battery residual life forecasting method of dynamic gray related vector machine |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4821962B2 (en) * | 2005-06-30 | 2011-11-24 | トヨタ自動車株式会社 | Fuel cell system |
-
2013
- 2013-07-25 CN CN201310317281.9A patent/CN103389471B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1813199A (en) * | 2003-06-25 | 2006-08-02 | 罗伯特.博世有限公司 | Method for predicting the residual service life of an electric energy accumulator |
| CN102262211A (en) * | 2011-04-20 | 2011-11-30 | 哈尔滨工业大学 | Analog circuit failure diagnosis method based on dynamic classification of echo state network |
| CN102798823A (en) * | 2012-06-15 | 2012-11-28 | 哈尔滨工业大学 | Gaussian process regression-based method for predicting state of health (SOH) of lithium batteries |
| CN103033761A (en) * | 2012-12-17 | 2013-04-10 | 哈尔滨工业大学 | Lithium ion battery residual life forecasting method of dynamic gray related vector machine |
Non-Patent Citations (2)
| Title |
|---|
| A Modified Echo State Network based Remaining Useful Life Estimation Approach;Yu Peng等;《Prognostics and Health Management Conference》;20120621;第1-7页 * |
| Data-driven Prognostics for Lithium-ion Battery Based on Gaussian Process Regression;Datong Liu等;《Prognostics and System Health Management Conference》;20120525;第1-5页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103389471A (en) | 2013-11-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103389471B (en) | A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section | |
| Park et al. | LSTM-based battery remaining useful life prediction with multi-channel charging profiles | |
| Xuan et al. | Real-time estimation of state-of-charge in lithium-ion batteries using improved central difference transform method | |
| Wang et al. | Improved anti-noise adaptive long short-term memory neural network modeling for the robust remaining useful life prediction of lithium-ion batteries | |
| Zhang et al. | An on-line estimation of battery pack parameters and state-of-charge using dual filters based on pack model | |
| Khalid et al. | Prediction of li-ion battery state of charge using multilayer perceptron and long short-term memory models | |
| CN106055775B (en) | A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model | |
| CN110568359B (en) | Lithium battery residual life prediction method | |
| CN103336913B (en) | A kind of method being suitable for space lithium ion battery status monitoring and cut-off voltage prediction | |
| CN108872866A (en) | A kind of charge states of lithium ion battery dynamic evaluation and long-acting prediction fusion method | |
| CN103439666B (en) | A kind of method of geometry of capacity of lithium ion battery slump evaluations | |
| CN103954915A (en) | Lithium ion battery remaining life indirect prediction method based on probability integration | |
| Boujoudar et al. | Lithium-ion batteries modeling and state of charge estimation using artificial neural network | |
| CN103983920B (en) | A kind of method of the model of the electrokinetic cell setting up electric vehicle | |
| CN105629175A (en) | Lithium ion battery life prediction method based on unscented Kalman filtering (UKF) | |
| CN104680024A (en) | Method for predicting remaining useful life of lithium ion battery based on GA (Genetic Algorithms) and ARMA (Auto Regressive and Moving Average) models | |
| CN114636932A (en) | Method and system for predicting remaining service life of battery | |
| CN107340476A (en) | The electrical state monitoring system and electrical state monitoring method of battery | |
| Mazzi et al. | State of charge estimation of an electric vehicle's battery using tiny neural network embedded on small microcontroller units | |
| Zhao et al. | Digital twin-driven estimation of state of charge for Li-ion battery | |
| Wen et al. | Lithium battery health state assessment based on vehicle-to-grid (V2G) real-world data and natural gradient boosting model | |
| CN111337833B (en) | Lithium battery capacity integrated prediction method based on dynamic time-varying weight | |
| Lipu et al. | Differential search optimized random forest regression algorithm for state of charge estimation in electric vehicle batteries | |
| CN115993537A (en) | Lithium battery capacity prediction method based on correlation analysis and WOA-LSTM | |
| Zhao et al. | Lithium-ion battery state-of-health estimation method using isobaric energy analysis and pso-lstm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |