CN114861545A - Lithium battery SOP online estimation method based on RNN neural network and multi-parameter constraint - Google Patents
Lithium battery SOP online estimation method based on RNN neural network and multi-parameter constraint Download PDFInfo
- Publication number
- CN114861545A CN114861545A CN202210553725.8A CN202210553725A CN114861545A CN 114861545 A CN114861545 A CN 114861545A CN 202210553725 A CN202210553725 A CN 202210553725A CN 114861545 A CN114861545 A CN 114861545A
- Authority
- CN
- China
- Prior art keywords
- soc
- battery
- lithium battery
- current
- 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.)
- Pending
Links
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 title claims abstract description 67
- 229910052744 lithium Inorganic materials 0.000 title claims abstract description 67
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 21
- 238000003062 neural network model Methods 0.000 claims abstract description 6
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 claims abstract description 4
- 229910001416 lithium ion Inorganic materials 0.000 claims abstract description 4
- 230000010287 polarization Effects 0.000 claims description 27
- 230000008569 process Effects 0.000 claims description 19
- 238000012360 testing method Methods 0.000 claims description 19
- 238000007599 discharging Methods 0.000 claims description 16
- 238000001914 filtration Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 230000002459 sustained effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000012886 linear function Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 239000003990 capacitor Substances 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 150000001875 compounds Chemical class 0.000 claims description 3
- 230000009897 systematic effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 5
- 210000004027 cell Anatomy 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000012897 Levenberg–Marquardt algorithm Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011088 calibration curve Methods 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000009194 climbing Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000295 fuel oil Substances 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/06—Power analysis or power optimisation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/70—Energy storage systems for electromobility, e.g. batteries
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Tests Of Electric Status Of Batteries (AREA)
Abstract
The invention discloses an RNN neural network and multi-parameter constraint-based lithium ion battery SOP online real-time estimation method. And based on the constraint of the SOC of the lithium battery and the continuous peak current of the circuit model, the multi-parameter constraint real-time estimation of the SOP of the lithium battery is completed by using an RNN neural network. According to the method, the deviation of the SOP of the lithium battery is smaller by estimating through the RNN neural network model and adding multi-parameter constraint, and the SOP of the battery is estimated by considering various constraint conditions, so that the estimation precision of the SOP is improved.
Description
Technical Field
The invention belongs to the field of battery management of electric vehicles, and particularly relates to an on-line estimation method for SOP of a lithium battery.
Background
With the continuous development of the automobile industry, the contradiction of energy consumption, environmental pollution and the like brought by the traditional fuel oil automobile is gradually highlighted, and becomes a factor which must be considered in the social progress. Therefore, the development of the electric automobile is enhanced, the use cost of the electric automobile is reduced, a low-carbon sustainable development strategy is implemented, and the worry of people about energy and environmental problems can be relieved.
The state of power (SOP) of a lithium battery generally refers to a sustained peak power state, which represents sustained peak discharge or charge power at different states of charge over a subsequent period of time, on the premise of ensuring normal operation of the battery. Accurate estimation of the SOP is of great significance in ensuring maximum recovery of braking energy during braking of the electric vehicle, providing as much power as possible during acceleration or starting, and shortening the time required for full charge as much as possible in the battery fast charge mode.
The estimation method for SOP mainly includes interpolation, modeling, and data-driven methods. Interpolation methods based on HPPC testing often require a large number of experiments to test the charge and discharge power values after the cell voltage reaches a defined voltage. Although the principle is simple, the method ignores the polarization phenomenon of the battery and has poor dynamic characteristics. The model method is further classified into a battery model-based method and an SOC-based method. The method based on the battery model can well represent the characteristic of the non-linearity of the SOP of the lithium battery, but different models can influence the estimation accuracy of the SOP. The method based on SOC alone causes the problem of potential safety hazard due to overlarge estimated peak current, and the method based on RNN neural network data driving has accurate input and output and high reference value.
Disclosure of Invention
In order to solve the technical problems mentioned in the background art, the invention provides an RNN neural network and multi-parameter constraint-based lithium battery SOP online estimation method.
In order to achieve the technical purpose, the technical scheme of the invention is as follows:
an RNN neural network and multi-parameter constraint-based lithium battery SOP online estimation method comprises the following steps:
(1) establishing a lithium battery equivalent circuit model and a dynamic system equation, performing an OCV test on the lithium battery, recording data of the lithium battery under a UDDS working condition, and performing online identification on each parameter of the lithium battery through an extended Kalman filtering algorithm; the parameter comprises an internal resistance R 0 A polarization resistor and a polarization capacitor;
(2) an EKF discrete nonlinear system equation is established according to a lithium battery equivalent circuit model, a state equation and an observation equation of a lithium battery are established through extended Kalman filtering, and online estimation of SOC under a circulation working condition is completed;
(3) estimating a sustained peak current and a sustained peak power;
(4) and (3) importing an RNN neural network model by taking terminal voltage, current and SOC (state of charge) of the battery under the working condition as input quantities and taking peak power SOP as output quantity, and establishing an RNN neural network to finish the estimation of the peak power of the lithium battery.
Further, in the step (1), the equivalent circuit model of the lithium battery is a second-order RC circuit model, and the second-order RC circuit model comprises electrochemical polarization internal resistance R 1 Electrochemical polarization capacitance C 1 Concentration polarization resistance R 2 Sum concentration polarization capacitance C 2 ；
The dynamic system equation is as follows:
U 0 ＝U oc (t)-R 0 ·I(t)-U 1 (t)-U 2 (t)
wherein, U 0 Representing the terminal voltage, U, of the battery in the operating state oc (t) is an open circuit voltage expression, R 0 Is an ohmic internal resistance expression, I represents the magnitude of the discharge current, U 1 And U 2 Respectively expressed as two groups of RC loop terminal voltages, and t is time.
Further, in the step (1), an OCV test is performed on the 18650 lithium ion battery at normal temperature, and test data are fitted to obtain a corresponding relation curve of the open-circuit voltage and the SOC.
Further, the specific process of step (1) is as follows:
(11) completing the UDDS working condition test of the lithium battery at normal temperature and environment temperature, and recording a voltage, current and an SOC true value;
(12) storing the intermediate process estimate;
(13) initializing parameters and converting the estimated values of the parameters to R 0 ,R 1 ,R 2 ,C 1 ,C 2 (ii) a Updating the next time value;
(14) calculating a Kalman filtering gain matrix, and acquiring voltage and current data at the next moment;
(15) and (5) repeating the steps (12) to (14) and identifying the parameters of the lithium battery model on line.
Further, the specific process of step (2) is as follows:
(21) establishing an EKF discrete nonlinear system equation:
χ k+1 ＝f(x k ,u k )+w k
y k ＝g(x k ,u k )+v k
wherein, χ k+1 To input, y k To output, w k And v k Is systematic noise and follows a normal distribution, with f (x) k ,u k )，g(x k ,u k ) Independently of one another, f (x) k ,u k ) And g (x) k ,u k ) Is a non-linear function;
(22) establishing a state equation and an observation equation of the lithium battery through an extended Kalman filtering algorithm:
U k ＝U oc (s k )-i k R 0 -U 1,k -U 2,k
wherein, tau d And τ e For two different time constants, U oc Is an open circuit voltage, s k Is an SOC estimate at time k, i k Current at time k, T sample time, C N Is the rated capacity of the battery, eta t For charging and discharging efficiency, U k Is a total estimate of the terminal voltage at time k, U 1,k Is at time k R 1 Estimated value of terminal voltage of U 2,k Is at time k R 2 A terminal voltage estimate of;
(23) and performing EKF recursive calculation on the discretized state equation and observation equation according to terminal voltage and current data acquired in the charging and discharging process of the lithium battery, and completing the online estimation of the SOC under the circulating working condition.
Further, the specific process of step (3) is as follows:
(31) based on the battery SOC, the peak current is estimated:
suppose a battery is operating at Δ t units of timeWith an internal current i k When discharging is performed, the battery SOC at the time t + Δ t is as follows:
when the battery works, the SOC value of the battery at any time meets the following constraint: SOC min <SOC(t)<SOC max Then, the maximum charge-discharge current under SOC constraint can be derived from the above equation:
in the formula eta i For cell coulombic efficiency, Q v For the actual capacity of the battery, SOC (t) is the SOC value at time t, SOC min 、SOC max Respectively the minimum and maximum values of SOC, eta chg And η dis The charging and discharging efficiency of the battery is improved,
for the minimum charging current based on the SOC of the battery,
is the maximum discharge current based on the battery SOC;
(32) estimating peak current based on a second-order RC model of the battery:
according to terminal voltage constraints: u shape t,min <U t <U t,max The maximum estimated charge-discharge current is obtained as follows:
in the formula (I), the compound is shown in the specification,
and
respectively estimating peak discharge and charging current at the moment k based on a second-order RC model of the battery, wherein L is continuous sampling time; g (SOC) represents the open circuit voltage of the battery, SOC k+1 Represents the SOC and the current i of the battery at the moment k +1 k+1 The functional relationship of (a);
g (SOC) first order partial derivatives of SOC;
(33) estimating the peak power of the lithium battery based on the multi-parameter constraint of the battery SOC and the battery second-order RC model:
first, the peak current under multi-parameter constraints is estimated:
wherein i max 、i min Respectively a preset maximum discharge current and a preset minimum charge current,
and
respectively obtaining the maximum value of the maximum discharging current and the minimum value of the minimum charging current based on the SOC of the battery in the continuous sampling time L;
the peak power is then estimated:
wherein, C N J represents the initial moment in the continuous sampling time, tau is the time constant,
is the maximum discharge peak power and minimum charge peak power, g (SOC, C) under multi-parameter constraints N ) Indicating the open circuit voltage of the battery based on the rated capacity of the battery,
denotes g (SOC, C) N ) And solving a first-order partial derivative of the SOC.
Further, the specific process of establishing the RNN neural network in step (4) is as follows:
(f) selecting an input and output variable;
(g) sample normalization processing;
(h) RNN network model selection;
(i) RNN network training;
(j) RNN network estimation.
Adopt the beneficial effect that above-mentioned technical scheme brought:
in the invention, a new RNN neural network model is adopted on the basis of the existing technology for estimating the SOP of the battery, and the multi-parameter constraint conditions of the SOC and the battery model are added to carry out new estimation on the SOP, so that the accuracy of the estimation model is greatly improved. Based on the invention, the utilization rate of the battery energy can be effectively improved, and the service life of the battery can be prolonged. The method has the advantages that key indexes of acceleration and climbing performance of the electric automobile are effectively evaluated, the limit capacity of the charge and discharge power of the power battery is represented, and the method is important for the motor to exert the energy recovery function to the maximum extent.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic diagram of an equivalent circuit model of a lithium battery according to the present invention;
FIG. 3 is a calibration curve diagram of OCV-SOC of the lithium battery in the embodiment;
FIG. 4 is an SOC estimation diagram of a lithium battery in the embodiment;
FIG. 5 shows the estimation result of the discharge peak power of the lithium battery in the embodiment;
FIG. 6 shows the estimation result of the peak charging power of the lithium battery in the embodiment;
Detailed Description
The technical scheme of the invention is explained in detail in the following with the accompanying drawings.
The invention designs a lithium battery SOP on-line estimation method based on RNN neural network and multi-parameter constraint, as shown in figure 1, the steps are as follows:
step 2, an EKF discrete nonlinear system equation is established according to a lithium battery equivalent circuit model, a state equation and an observation equation of the lithium battery are established through extended Kalman filtering, and online estimation of SOC under a circulation working condition is completed;
estimating continuous peak current by adopting a method based on an SOC (system on chip) and a method based on a battery model, and estimating continuous peak power by adopting a method based on multi-parameter constraint of the SOC and the battery model;
and 4, importing an RNN neural network model by using terminal voltage, current and SOC (state of charge) under the working condition of the battery as input quantities and using peak power SOP as output quantities, establishing an RNN neural network, and finishing the estimation of the peak power of the lithium battery.
In this embodiment, in step 1, the equivalent circuit model of the lithium battery adopts a second-order RC circuit model, as shown in fig. 2, the second-order RC circuit model includes an electrochemical polarization internal resistance, an electrochemical polarization capacitance, a concentration polarization resistance, and a concentration polarization capacitance. U shape oc Indicating the open circuit voltage, U, of the battery in the inoperative state o Representing the terminal voltage of the battery in the working state; i represents the magnitude of the discharge current; r 0 Indicating the internal ohmic internal resistance, R, of the battery 1 ,R 2 ,C 1 ,C 2 Two groups of polarization parameters are respectively expressed as electrochemical polarization internal resistance, concentration polarization internal resistance, electrochemical polarization capacitance and concentration polarization capacitance, and the two groups of parameters can be respectively used for representing the quick and slow response characteristics in the internal reaction process of the battery, U 1 And U 2 Respectively, the terminal voltages of two groups of RC loops, and the sum of the terminal voltages is the total voltage of the battery.
In this embodiment, in step 1, if a dynamic system equation of the battery is to be established, counting needs to be performed with reference to the SOC of the ampere-hour integration method, and a corresponding second-order model state equation of the battery may be described as:
wherein, SOC (0) is the SOC value of the battery in the initial state, and SOC (t) is the SOC value of the battery at the time t;
is the battery capacity at a particular discharge rate; eta is the coulomb coefficient of the cell; i (t) is the current at time t; u shape 1 (t)，U 2 (t) terminal voltages of two RC combinations at time t; r 1 ,C 1 ,R 2 ,C 2 Four polarization parameters.
The battery dynamic system equation can be described as:
U 0 ＝U oc (t)-R 0 ·I(t)-U 1 (t)-U 2 (t)
wherein, U oc (t) is the open circuit voltage at time t, R 0 Is an ohmic internal resistance expression. I (t) represents the discharge current at time t, U 1 (t) and U 2 And (t) respectively represents two groups of RC loop terminal voltages at the time t.
In this embodiment, in step 1, an OCV test is performed at normal temperature and ambient temperature using a 18650 lithium battery, and the test data is fitted to obtain a corresponding relationship curve between the open-circuit voltage and the SOC. The specific process is as follows:
1a, fully standing the lithium battery, and then carrying out constant current discharge with a current of 0.3A. After the battery was fully left at the cut-off voltage (2.8V) and the SOC was 0, the battery was charged at a constant current and a constant voltage of 0.3A and a cut-off current of 0.03A after the battery was left at rest for 2 hours, and at the cut-off voltage (4.2V) and the SOC was 100% after the charge current was 0.03A or less, the battery was left at rest for 2 hours. The OCV test was then started, first left for 10s, followed by constant current discharge for 2h, and then left for 2h, at which point the battery SOC was considered to have dropped to 90%, and the cycle was ten times until the battery SOC was 0, completing the OCV test.
1b, performing curve fitting on the open-circuit voltage value corresponding to the time point when the SOC value is from 100% to 10%, and obtaining a relation curve graph of the open-circuit voltage and the SOC, wherein the relation curve graph is shown in FIG. 3.
In this embodiment, the specific process of step 1 is as follows:
1a, regarding the lithium battery as a system, the state equation in the system can be expressed as:
x k+1 ＝f(x k ,u k )+w k y k ＝g(x k ,u k )+v k
in the formula, x k Is a state variable; u. of k Is the excitation of the system; w is a k ，v k Is the system noise; y is k Is the output quantity of the system; f (x) k ,u k ) And g (x) k ,u k ) Is a non-linear function.
1b, transforming the equivalent circuit state equation into: u shape O ＝U OC -U 1 -U 2 -R 0 I。
1c, if the relative change of the battery open-circuit voltage is zero in a short time, the time derivative is obtained by using the battery output voltage:
wherein, U o Is terminal voltage value, t is time, R 1 And C 1 ，R 2 And C 2 Two groups of electrochemical polarization internal resistance and concentration polarization capacitance, R 0 Is internal resistance, U 1 Is a set of polarized internal resistance and voltage on the concentration polarized capacitor, I is current value, U oc Is an open circuit voltage.
1d, converting the above formula into a state equation, wherein the system state variable of the battery equivalent circuit is as follows:
x＝[U 1 U 2 1/R 1 1/R 2 1/C 1 1/C 2 ]
wherein, the battery input is: u. of k I wherein u k Is the battery terminal voltage at time k.
And 1e, according to the corresponding relation of the input current and output voltage response curves of the lithium battery, obtaining the relation between the state variable of the battery system and the unknown parameters in the system by utilizing the above formula to obtain each parameter under different SOC.
And 1f, completing the UDDS working condition to obtain a parameter online identification result.
In this embodiment, the specific process of step 2 is as follows:
2a, establishing an EKF discrete nonlinear system equation:
χ k+1 ＝f(x k ,u k )+w k
y k ＝g(x l ,u k )+v k
wherein, χ l+1 To input, y k To output, w k And v k Is systematic noise and follows a normal distribution, with f (x) k ,u k )，g(x k ,u k ) Independently of one another, f (x) k ,u k ) And g (x) k ,u k ) Is a non-linear function.
2b, establishing a state equation and an observation equation of the lithium battery through an extended Kalman filtering algorithm:
U k ＝U oc (s k )-i k R 0 -U 1,k -U 2,k
wherein, tau d And τ e For two different time constants, U oc Is an open circuit voltage, s k Is an SOC estimate at time k, i k Current at time k, T sample time, C N Is the rated capacity of the battery, eta t For charging and discharging efficiency, U k Terminal voltage at time kTotal estimate, U 1,k Is at time k R 1 Estimated value of terminal voltage of U 2,k Is at time k R 2 A terminal voltage estimate of;
and 2c, performing EKF recursive calculation on the discretized state equation and observation equation according to the collected terminal voltage and current data in the lithium battery charging and discharging process, and completing the online estimation of the SOC under the circulating working condition.
Firstly, initializing a system:
the current value I at the current moment, the SOC value at the current moment and the terminal voltage value U at the current moment are measured 0 As input, substituting into extended Kalman filter algorithm, and inputting battery rated capacity C N (ii) a The sample time T and the total step size N are input.
Calculating an error covariance matrix:
immediately after the state prediction is calculated:
updating Kalman filtering gain:
correcting the state predicted value:
correcting the covariance matrix:
P k/k ＝(I-H k C k )P k/k-1
and obtaining a current value and a voltage value at the next moment, and estimating the SOC online.
In this embodiment, the specific process of step 3 is as follows:
and 3a, calculating the maximum current value which can be borne by the battery under the current condition according to the limitation of the SOC, and multiplying the maximum current value by the terminal voltage of the corresponding battery to calculate the maximum charge-discharge power. Assuming constant current discharge of the battery in delta t unit time, the current value is i k Then, the battery SOC at time t + Δ t may be expressed as:
in the formula eta i For cell coulombic efficiency, Q v For the actual capacity of the battery, when the battery works, the SOC value is required to be satisfied in the following interval at any time: SOC min <SOC(t)<SOC max Then, the maximum charge-discharge current under SOC constraint can be derived from the above equation:
η chg and η dis In order to achieve the charge-discharge efficiency of the battery,
in order to minimize the charging current,
is the maximum discharge current. SOC (t) is the SOC value at time t, as shown in FIG. 4.
It can be seen from the above derivation process that the method based on the battery SOC limit can calculate the continuous peak power of the battery in unit time Δ t, essentially to ensure the safety of the power battery, so the estimated value has a certain margin. Peak current calculated only with battery SOC as a constraint will also bias the peak power estimation result to a large value.
Peak current estimation based on second order RC model
According to the terminal voltage range: u shape t,min <U t <U t,max The maximum available charge-discharge current estimate is:
in the formula (I), the compound is shown in the specification,
and
the estimates of peak discharge and charge currents at time k based on a battery second order RC model, respectively. L is the duration sample time. g (SOC) represents the open circuit voltage of the battery, SOC k+1 Represents the SOC and the current i of the battery at the moment k +1 k+1 The functional relationship of (a); q v Which represents the actual capacity of the battery,
g (SOC) first order partial derivatives of SOC. U shape 1,k Is at time k R 1 Estimated value of terminal voltage of U 2,k Is at time k R 2 The terminal voltage estimate of (c).
Continuous peak power estimation under multi-parameter constraints based on SOC and battery model, as shown in fig. 5 and 6:
and solving the dynamic peak current which simultaneously meets the dynamic voltage characteristics, current limit value and SOC constraint conditions of the lithium battery:
wherein i max 、i min Respectively a maximum discharge current and a minimum charge current under assumed conditions,
the maximum continuous discharge current and the minimum continuous charge current under the constraint condition of the battery model are respectively.
And
and respectively obtaining the estimation values of the continuous maximum charging and discharging current (taking the maximum value of the maximum discharging current and the minimum value of the minimum charging current based on the battery SOC in the continuous sampling time L) based on the constraint of the battery SOC in the current duration.
The peak power is then estimated:
wherein, C N For the rated capacity of the battery, L is a continuous sampling time, j represents an initial moment in the continuous sampling time, tau is a time constant, and delta t is a unit time. g (SOC, C) N ) Shows the open-circuit voltage of the battery based on the rated capacity of the battery, SOC (k) is the SOC at the time k,
denotes g (SOC, C) N ) And solving a first-order partial derivative of the SOC. U shape 1,k Is at time k R 1 Estimated value of terminal voltage of U 2,k Is at time k R 2 The terminal voltage estimate of (c).
Is the maximum discharge peak power and the minimum charge peak power under the current constraints.
In this embodiment, the specific process of step 4 is as follows:
4a, selecting input and output variables
The accuracy, the calculation amount and the like of the neural network estimation model are closely related to the selection of the input variables and the quantity of the variables. In order to improve the estimation accuracy of the RNN network, factors having a large influence on the maximum charge/discharge power of the lithium ion battery should be selected as much as possible, but the requirements of the network scale and the calculation amount should be considered at the same time. The current, terminal voltage and state of charge (SOC) of the battery under the UDDS working condition are selected as input variables of the RNN network peak power estimation model.
4b. sample normalization processing
After the input variables are selected, training samples required by the RNN model need to be obtained, and normalization processing is carried out on the sample data. For battery input real-time data acquisition, 2000 groups of samples are selected, 1900 groups of samples are randomly selected as network training sample data, and the remaining 100 groups of data are used as network estimation verification data.
RNN network model selection
Because the number of input variables of the RNN network model for power estimation built by the method is 4, the number of nodes of a network input layer is set to be 4, the number of nodes of a hidden layer is set to be 5, a transfer function from the input layer to the hidden layer is selected to be a purelin linear function by the node 1 of the output layer, the transfer function from the hidden layer to the output layer is a tan sign function, and a network learning algorithm selects a tranlm or Levenberg-Marquardt algorithm; and selecting the RNN learning rule as the learngdm for updating the network weight and the threshold.
RNN network training
And importing the normalized training sample data into an input layer of the RNN, and then training the network. The connection weight values and the network threshold values among the neurons of each layer are set by default, and parameters such as network learning rate, maximum iteration times, precision and the like can be initialized according to the following formula:
net.trainParam.show＝50
net.trainParam.lr＝0.09
net.trainParam.epochs＝1000
net.trainParam.goal＝0.000001
net.trainParam.min_grad＝le-12
wherein net, trian param, show indicates that the network displays the training result every 50 times of training; net. trainparam. lr means that the network learning rate is 0.09(ii) a net, trainparam, epochs indicate that the maximum number of iterations of the network loop is 1000; net, trainparam, epochs, indicates that the sum of the squares of the errors of the estimated output and the expected output of the network is 0.000001; and stopping training when the total error reaches 0.000004; net, trainParam, min _ grad represents a minimum value of the gradient of the performance index function of 10 -12 . The network is then trained.
RNN network estimation
Normalization of the estimated sample data:
inputn_test＝map minmax(′apply′,input_test,inputps)
and (3) testing the estimation sample by using the trained RNN model:
an＝sim(net,inputn_test)
the network estimation output is inversely normalized:
Bpoutput＝mapminmax(′reverse′,an,outputps)
according to the steps, estimating the peak power of the lithium battery in the MATLAB environment. The results are as follows: the charge and discharge peak power estimation result of the lithium battery is obtained, and the following results can be obtained: the estimation errors of the charging and discharging peak power are all within 2%, the average error of the charging peak power is 1.76%, and the average error of the discharging peak power is 1.69%, so that the accuracy of estimating the SOP of the lithium battery by adopting an RNN neural network model is very high.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can understand that the modifications or substitutions within the technical scope of the present invention are included in the scope of the present invention, and therefore, the scope of the present invention should be subject to the protection scope of the claims.
Claims (7)
1. An RNN neural network and multi-parameter constraint-based lithium battery SOP online estimation method is characterized by comprising the following steps:
(1) establishing a lithium battery equivalent circuit model and a dynamic system equation, carrying out an OCV test on the lithium battery, recording data of the lithium battery under the UDDS working condition, and carrying out an extended Kalman filtering algorithm on the dataIdentifying each parameter of the lithium battery on line; the parameter comprises an internal resistance R 0 A polarization resistor and a polarization capacitor;
(2) establishing an EKF discrete nonlinear system equation according to the lithium battery equivalent circuit model, establishing a state equation and an observation equation of the lithium battery through extended Kalman filtering, and completing online estimation of SOC under the circulation working condition;
(3) estimating a sustained peak current and a sustained peak power;
(4) and (3) importing an RNN neural network model by taking terminal voltage, current and SOC (state of charge) of the battery under the working condition as input quantities and taking peak power SOP as output quantity, and establishing an RNN neural network to finish the estimation of the peak power of the lithium battery.
2. The lithium battery SOP online estimation method based on the RNN neural network and the multi-parameter constraint as claimed in claim 1, wherein in the step (1), the lithium battery equivalent circuit model is a second-order RC circuit model, and the second-order RC circuit model comprises electrochemical polarization internal resistance R 1 Electrochemical polarization capacitance C 1 Concentration polarization resistance R 2 Sum concentration polarization capacitance C 2 ；
The dynamic system equation is as follows:
U 0 ＝U oc (t)-R 0 ·I(t)-U 1 (t)-U 2 (t)
wherein, U 0 Representing the terminal voltage, U, of the battery in the operating state oc (t) is an open circuit voltage expression, R 0 Is an ohmic internal resistance expression, I represents the magnitude of the discharge current, U 1 And U 2 Respectively expressed as two groups of RC loop terminal voltages, and t is time.
3. The lithium battery SOP online estimation method based on the RNN neural network and the multi-parameter constraint as claimed in claim 2, wherein in the step (1), an OCV test is performed at normal temperature by adopting a 18650 lithium ion battery, and a corresponding relation curve of open-circuit voltage and SOC is obtained by fitting test data.
4. The lithium battery SOP on-line estimation method based on the RNN neural network and the multi-parameter constraint as claimed in claim 2, wherein the specific process of the step (1) is as follows:
(11) completing the UDDS working condition test of the lithium battery at normal temperature and environment temperature, and recording a voltage, current and an SOC true value;
(12) storing the intermediate process estimate;
(13) initializing parameters and converting the parameter estimates to R 0 ,R 1 ,R 2 ,C 1 ,C 2 (ii) a Updating the next time value;
(14) calculating a Kalman filtering gain matrix, and acquiring voltage and current data at the next moment;
(15) and (5) repeating the steps (12) to (14) and identifying the parameters of the lithium battery model on line.
5. The lithium battery SOP on-line estimation method based on the RNN neural network and the multi-parameter constraint as claimed in claim 4, wherein the specific process of the step (2) is as follows:
(21) establishing an EKF discrete nonlinear system equation:
χ k+1 ＝f(x k ,u k )+w k
y k ＝g(x k ,u k )+v k
wherein, χ k+1 To input, y k To output, w k And v k Is systematic noise and follows a normal distribution, with f (x) k ,u k )，g(x k ,u k ) Independently of one another, f (x) k ,u k ) And g (x) k ,u k ) Is a non-linear function;
(22) establishing a state equation and an observation equation of the lithium battery through an extended Kalman filtering algorithm:
U k ＝U oc (s k )-i k R 0 -U 1,k -U 2,k
wherein, tau d And τ e For two different time constants, U oc Is an open circuit voltage, s k Is an SOC estimate at time k, i k Current at time k, T sample time, C N Is the rated capacity of the battery, eta t For charging and discharging efficiency, U k Is a total estimate of the terminal voltage at time k, U 1,k Is at time k R 1 Estimated value of terminal voltage of U 2,k Is at time k R 2 A terminal voltage estimate of;
(23) and performing EKF recursive calculation on the discretized state equation and observation equation according to terminal voltage and current data acquired in the charging and discharging process of the lithium battery, and completing the online estimation of the SOC under the circulating working condition.
6. The lithium battery SOP on-line estimation method based on the RNN neural network and the multi-parameter constraint according to claim 5, wherein the specific process of the step (3) is as follows:
(31) based on the battery SOC, the peak current is estimated:
suppose a battery is charged with a current i during a unit time of Δ t k When discharging is performed, the battery SOC at the time t + Δ t is as follows:
when the battery works, the SOC value of the battery at any time meets the following constraint: SOC min <SOC(t)<SOC max Then, the maximum charge-discharge current under SOC constraint can be derived from the above equation:
in the formula eta i For cell coulombic efficiency, Q v For the actual capacity of the battery, SOC (t) is the SOC value at time t, SOC min 、SOC max Respectively the minimum and maximum values of SOC, eta chg And η dis The charging and discharging efficiency of the battery is improved,
for the minimum charging current based on the SOC of the battery,
is the maximum discharge current based on the battery SOC;
(32) estimating peak current based on a second-order RC model of the battery:
according to terminal voltage constraints: u shape t,min <U t <U t,max The maximum estimated charge-discharge current is obtained as follows:
in the formula (I), the compound is shown in the specification,
and
respectively estimating peak discharge and charging current at the moment k based on a second-order RC model of the battery, wherein L is continuous sampling time; g (SOC) represents the open circuit voltage of the battery, SOC k+1 Represents the SOC and the current i of the battery at the moment k +1 k+1 The functional relationship of (a);
g (SOC) first order partial derivatives of SOC;
(33) estimating the peak power of the lithium battery based on the multi-parameter constraint of the battery SOC and the battery second-order RC model:
first, the peak current under multi-parameter constraints is estimated:
wherein i max 、i min Respectively a preset maximum discharge current and a preset minimum charge current,
and
respectively obtaining the maximum value of the maximum discharging current and the minimum value of the minimum charging current based on the SOC of the battery in the continuous sampling time L;
the peak power is then estimated:
wherein, C N J represents the initial moment in the continuous sampling time, tau is the time constant,
is the maximum discharge peak power and minimum charge peak power, g (SOC, C) under multi-parameter constraints N ) Indicating the open circuit voltage of the battery based on the rated capacity of the battery,
denotes g (SOC, C) N ) And solving a first-order partial derivative of the SOC.
7. The lithium battery SOP on-line estimation method based on the RNN neural network and the multi-parameter constraint as claimed in claim 1, wherein the specific process of establishing the RNN neural network in the step (4) is as follows:
(a) selecting an input and output variable;
(b) sample normalization processing;
(c) RNN network model selection;
(d) RNN network training;
(e) RNN network estimation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210553725.8A CN114861545A (en) | 2022-05-19 | 2022-05-19 | Lithium battery SOP online estimation method based on RNN neural network and multi-parameter constraint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210553725.8A CN114861545A (en) | 2022-05-19 | 2022-05-19 | Lithium battery SOP online estimation method based on RNN neural network and multi-parameter constraint |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114861545A true CN114861545A (en) | 2022-08-05 |
Family
ID=82639222
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210553725.8A Pending CN114861545A (en) | 2022-05-19 | 2022-05-19 | Lithium battery SOP online estimation method based on RNN neural network and multi-parameter constraint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114861545A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116449221A (en) * | 2023-06-14 | 2023-07-18 | 浙江天能新材料有限公司 | Lithium battery state of charge prediction method, device, equipment and storage medium |
| CN116500457A (en) * | 2023-06-26 | 2023-07-28 | 西北工业大学 | Method for estimating battery SOC by fusing neural network with RC equivalent circuit model |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015188610A1 (en) * | 2014-06-11 | 2015-12-17 | 北京交通大学 | Method and device for estimating state of charge of battery |
| CN111352032A (en) * | 2018-12-21 | 2020-06-30 | 重庆邮电大学 | Lithium battery dynamic peak power prediction method |
| CN113608126A (en) * | 2021-09-07 | 2021-11-05 | 南京邮电大学 | Lithium battery SOC online estimation method at different temperatures |
-
2022
- 2022-05-19 CN CN202210553725.8A patent/CN114861545A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015188610A1 (en) * | 2014-06-11 | 2015-12-17 | 北京交通大学 | Method and device for estimating state of charge of battery |
| CN111352032A (en) * | 2018-12-21 | 2020-06-30 | 重庆邮电大学 | Lithium battery dynamic peak power prediction method |
| CN113608126A (en) * | 2021-09-07 | 2021-11-05 | 南京邮电大学 | Lithium battery SOC online estimation method at different temperatures |
Non-Patent Citations (1)
| Title |
|---|
| 韩忠华;刘珊珊;石刚;董挺;: "基于扩展卡尔曼神经网络算法估计电池SOC", 电子技术应用, no. 07, 12 July 2016 (2016-07-12) * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116449221A (en) * | 2023-06-14 | 2023-07-18 | 浙江天能新材料有限公司 | Lithium battery state of charge prediction method, device, equipment and storage medium |
| CN116449221B (en) * | 2023-06-14 | 2023-09-29 | 浙江天能新材料有限公司 | Lithium battery state of charge prediction method, device, equipment and storage medium |
| CN116500457A (en) * | 2023-06-26 | 2023-07-28 | 西北工业大学 | Method for estimating battery SOC by fusing neural network with RC equivalent circuit model |
| CN116500457B (en) * | 2023-06-26 | 2023-09-19 | 西北工业大学 | Method for estimating battery SOC by fusing neural network with RC equivalent circuit model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lin et al. | State of charge estimation with the adaptive unscented Kalman filter based on an accurate equivalent circuit model | |
| CN112946499B (en) | Lithium battery health state and state of charge joint estimation method based on machine learning | |
| Lai et al. | Co-estimation of state of charge and state of power for lithium-ion batteries based on fractional variable-order model | |
| Wang et al. | Probability based remaining capacity estimation using data-driven and neural network model | |
| CN110286332B (en) | Electric vehicle power battery SOC estimation method based on multiple innovation theory | |
| CN111323719A (en) | Method and system for online determination of health state of power battery pack of electric automobile | |
| CN111722118B (en) | Lithium ion battery SOC estimation method based on SOC-OCV optimization curve | |
| Ma et al. | Co-estimation of state of charge and state of health for lithium-ion batteries based on fractional-order model with multi-innovations unscented Kalman filter method | |
| CN110824363B (en) | Lithium battery SOC and SOE joint estimation method based on improved CKF | |
| CN105974327A (en) | Lithium battery pack SOC prediction method based on neural network and UKF | |
| CN110346734A (en) | A kind of lithium-ion-power cell health status evaluation method based on machine learning | |
| CN109358293B (en) | Lithium ion battery SOC estimation method based on IPF | |
| CN111913109B (en) | Method and device for predicting peak power of battery | |
| CN114861545A (en) | Lithium battery SOP online estimation method based on RNN neural network and multi-parameter constraint | |
| CN112989690B (en) | Multi-time-scale state-of-charge estimation method for lithium battery of hybrid electric vehicle | |
| CN113608126B (en) | Lithium battery SOC online prediction method under different temperatures | |
| CN112269133B (en) | SOC estimation method based on pre-charging circuit model parameter identification | |
| CN111983457A (en) | Battery pack SOH estimation method based on LSTM neural network | |
| Xia et al. | Joint estimation of the state-of-energy and state-of-charge of lithium-ion batteries under a wide temperature range based on the fusion modeling and online parameter prediction | |
| CN111812515A (en) | XGboost model-based lithium ion battery state of charge estimation | |
| CN109901072B (en) | Retired battery parameter detection method based on historical data and laboratory test data | |
| CN112098849A (en) | Lithium battery residual capacity estimation method based on integral Kalman filtering | |
| CN112858916A (en) | Battery pack state of charge estimation method based on model and data drive fusion | |
| CN112946481A (en) | Based on federation H∞Filtering sliding-mode observer lithium ion battery SOC estimation method and battery management system | |
| CN116125278A (en) | Lithium battery SOC estimation method and system based on LSTM-EKF algorithm |
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 |