CN113933725A - Method for determining power battery charge state based on data driving - Google Patents

Method for determining power battery charge state based on data driving Download PDF

Info

Publication number
CN113933725A
CN113933725A CN202111051231.1A CN202111051231A CN113933725A CN 113933725 A CN113933725 A CN 113933725A CN 202111051231 A CN202111051231 A CN 202111051231A CN 113933725 A CN113933725 A CN 113933725A
Authority
CN
China
Prior art keywords
data
battery
voltage
charge
state
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202111051231.1A
Other languages
Chinese (zh)
Other versions
CN113933725B (en
Inventor
李晓宇
徐健华
王腾远
田劲东
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen University
Original Assignee
Shenzhen University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen University filed Critical Shenzhen University
Priority to CN202111051231.1A priority Critical patent/CN113933725B/en
Publication of CN113933725A publication Critical patent/CN113933725A/en
Application granted granted Critical
Publication of CN113933725B publication Critical patent/CN113933725B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/385—Arrangements for measuring battery or accumulator variables
    • G01R31/387—Determining ampere-hour charge capacity or SoC
    • G01R31/388—Determining ampere-hour charge capacity or SoC involving voltage measurements
    • 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

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Secondary Cells (AREA)

Abstract

The embodiment of the application provides a method for determining the state of charge of a power battery based on data driving, which can be applied to various scenes and is suitable for both under-sampling conditions and non-under-sampling conditions.

Description

Method for determining power battery charge state based on data driving
Technical Field
The application belongs to the technical field of battery state monitoring, and particularly relates to a method for determining the state of charge of a power battery based on data driving.
Background
The power battery is a core component of the electric automobile, and in order to ensure the safety of the electric automobile, it is very important to accurately determine the state of the power battery. The state of charge (SOC) reflects the remaining available battery capacity, and is an important parameter for measuring the state of the power battery.
One current method for determining the SOC of a power battery is a data-driven method, in which data such as voltage, current, and temperature of the power battery under specific driving conditions are acquired by a sensor on the power battery, and then the data such as voltage, current, and temperature are input into a neural network model, and the SOC of the battery is output by the neural network model. However, under different driving conditions, such as different speeds or different temperatures, the voltage and the current corresponding to the same SOC value of the electric vehicle are different, and have strong volatility and randomness, and the driving conditions that can be simulated when training the neural network model are limited. At this time, it may happen that the actual driving condition is a driving condition that is not simulated when the neural network model is trained, and the SOC accuracy of the power battery obtained under the actual driving condition is low.
Disclosure of Invention
In view of the above technical problems, embodiments of the present application provide a method for determining a state of charge of a power battery based on data driving, so as to reduce volatility and randomness of data and improve accuracy of determining an SOC of the power battery.
In a first aspect, an embodiment of the present application provides a method for determining a state of charge of a power battery based on data driving, where the method includes:
acquiring first data of a battery, wherein the first data comprises a plurality of voltage data of the battery;
processing the first data to obtain second data, wherein the second data comprises first characteristics, and the first characteristics reflect the change trend of the plurality of voltage data;
and determining the state of charge of the battery according to the second data.
With reference to the first aspect, in some implementations of the first aspect, the processing the first data to obtain second data includes: and processing the first data according to an empirical mode decomposition method to obtain second data.
With reference to the first aspect and the foregoing implementation manners, in some implementation manners of the first aspect, after obtaining the second data, the method further includes: acquiring the internal resistance of the battery; correcting the second data according to the internal resistance of the battery to obtain corrected second data;
determining a state of charge of the battery based on the second data, comprising: and determining the state of charge of the battery according to the corrected second data.
The second data is corrected, so that the randomness and the fluctuation of the voltage data of the electric automobile under different driving conditions can be further reduced, and the accuracy of determining the SOC of the battery is improved.
With reference to the first aspect and the foregoing implementation manners, in some implementation manners of the first aspect, determining a state of charge of the battery according to the corrected second data includes: and inputting the corrected second data into a neural network model to obtain the state of charge of the battery, wherein the neural network model is used for outputting the state of charge of the battery according to the first characteristic.
With reference to the first aspect and the foregoing implementation manners, in certain implementation manners of the first aspect, the second data further includes a second feature, and the second feature is a high-frequency component of the plurality of voltage data.
With reference to the first aspect and the foregoing implementations, in certain implementations of the first aspect, the first data further includes a plurality of current data and a plurality of temperature data of the battery.
With reference to the first aspect and the foregoing implementation manners, in some implementation manners of the first aspect, the second data further includes a third feature, a fourth feature and temperature data, the third feature reflects a variation trend of the plurality of current data, and the fourth feature is a high-frequency component of the plurality of current data.
In a second aspect, an embodiment of the present application provides an apparatus for determining a state of charge of a power battery based on data driving, including:
the battery management system comprises an acquisition unit, a storage unit and a control unit, wherein the acquisition unit is used for acquiring first data of a battery, and the first data comprises a plurality of voltage data of the battery;
the processing unit is used for processing the first data to obtain second data, wherein the second data comprises first characteristics, and the first characteristics reflect the change trend of the plurality of voltage data; and determining the state of charge of the battery according to the second data.
In a third aspect, an embodiment of the present application further provides an electronic device, which includes a memory and a processor, where the memory stores a computer program, and the processor implements the method according to the first aspect when executing the computer program.
In a fourth aspect, embodiments of the present application further provide a computer-readable storage medium storing computer instructions that, when executed on a computer, cause the computer to perform the method according to the first aspect.
In a fifth aspect, the present application further provides a computer program product, which includes a computer program and implements the method according to the first aspect when the computer program product runs on a computer.
According to the method for determining the state of charge of the power battery based on data driving, the first data of the battery are obtained, the first data comprise a plurality of voltage data of the battery, the first data are processed, the first characteristics reflecting the change trend of the plurality of voltage data are extracted from the complex voltage data, the second data are obtained, the state of charge of the battery is determined according to the second data, the randomness and the fluctuation of the plurality of voltage data can be reduced, and the accuracy of determining the SOC of the battery is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
FIG. 1 is a schematic flow chart diagram illustrating a method for determining a state of charge of a power battery based on data driving according to an embodiment of the present disclosure;
FIG. 2 is a basic structure of an LSTM unit provided in an embodiment of the present application;
FIG. 3 is a neural network framework for SOC estimation provided by embodiments of the present application;
figure 4 is a data plot of the voltage, current and temperature of the cell at ambient temperatures-10 c and 25 c for the US06 operating conditions.
FIG. 5(a) is a pre-compensation voltage residual component for 60% SOC at-10 deg.C ambient temperature, Cycle 1, NN, UDDS, LA92, HWFET, and US 06;
FIG. 5(b) is the pre-compensation voltage residual component for 50% SOC at-10 deg.C ambient temperature, Cycle 1, NN, UDDS, LA92, HWFET, and US 06;
FIG. 5(c) is a pre-compensation voltage residual component for 40% SOC at-10 deg.C ambient temperature, Cycle 1, NN, UDDS, LA92, HWFET, and US 06;
FIG. 5(d) is the compensated voltage residual component for 60% SOC at-10 deg.C for Cycle 1, NN, UDDS, LA92, HWFET, and US 06;
FIG. 5(e) is the compensated voltage residual component for 50% SOC at-10 deg.C for Cycle 1, NN, UDDS, LA92, HWFET, and US06 operating conditions;
FIG. 5(f) is the compensated voltage residual component for 40% SOC at ambient temperature-10 deg.C, Cycle 1, NN, UDDS, LA92, HWFET, and US06 operating conditions.
FIG. 6(a) is a SOC estimation of HWFET operating conditions at ambient temperature-10 ℃ using two methods;
FIG. 6(b) is a SOC estimation of HWFET operating conditions at 25 deg.C ambient temperature using two methods;
FIG. 6(c) is the SOC error results for the HWFET operating conditions at ambient temperature-10 deg.C using both methods;
FIG. 6(d) is a SOC error result for HWFET operating conditions at ambient temperature of 25 deg.C using both methods;
FIG. 7(a) is a SOC estimation of the US06 operating condition at-10 ℃ ambient temperature using two methods;
FIG. 7(b) is a SOC estimation of the US06 operating condition at 25 deg.C ambient temperature using two methods;
FIG. 7(c) is the SOC error results for the US06 operating condition at ambient temperature-10 ℃ when both methods are used;
FIG. 7(d) is a SOC error result for the US06 operating condition at 25 deg.C ambient temperature using both methods;
FIG. 8 is a schematic diagram of an apparatus according to an embodiment of the present disclosure;
fig. 9 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
In the following, the terms "first", "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of the present application, "a plurality" means two or more, and "at least one", "one or more" means one, two or more, unless otherwise specified.
Reference throughout this specification to "one embodiment" or "some embodiments," or the like, means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the present application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," or the like, in various places throughout this specification are not necessarily all referring to the same embodiment, but rather "one or more but not all embodiments" unless specifically stated otherwise. The terms "comprising," "including," "having," and variations thereof mean "including, but not limited to," unless expressly specified otherwise.
The battery is used as a highly nonlinear and time-varying system, and internal parameters of the battery dynamically vary, so that the SOC of the power battery is difficult to accurately estimate. At present, methods for determining the SOC of a power battery are mainly classified into the following four methods: coulomb counting, Open Circuit Voltage (OCV), model-based and data-driven methods. The accuracy of coulomb counting is affected by the initial SOC error and the sensor accuracy, and accumulated errors are easily formed during long-term use. The open-circuit voltage method needs to keep the power battery for a period of time to obtain an accurate OCV, and obtain a corresponding SOC by looking up a table, however, keeping the power battery for a period of time during the actual operation of the electric vehicle is difficult to achieve.
The method based on the model realizes dynamic tracking of the characteristics of the power battery through the model, models the power battery by combining a Shepherd model, an Unnewerh Universal model and a Nernst model, adopts a recursive least square method to perform online parameter identification, determines the SOC by using unscented Kalman filtering, judges whether the SOC is abnormal by using a detection method based on residual error information characteristics, and introduces an adaptive attenuation factor to correct a covariance matrix reflecting errors. The method has the disadvantages that when the battery data is in an under-sampling state, systematic errors are introduced in the discretization process of model establishment and SOC determination.
The state of charge of the power battery is determined through a neural network model based on a data-driven method. Firstly, collecting working condition data of a battery, wherein the working condition data comprises voltage, current, temperature and current electric quantity, calculating an SOC training label of a neural network model according to the current electric quantity and standard electric quantity, normalizing the voltage, the current and the temperature, and then training by adopting a bidirectional LSTM Encoder-Decoder structure to obtain the neural network model. When the SOC is determined, the data such as the voltage, the current, the temperature and the like of the power battery under specific driving conditions are acquired through a sensor on the power battery, then the data such as the voltage, the current, the temperature and the like are input into a neural network model, and the SOC of the battery is output through the neural network model.
However, under different driving conditions (i.e. different working conditions), such as different speeds or different temperatures, the voltages and currents corresponding to the same SOC value are different, and have strong volatility and randomness, and under low temperature or high power, the volatility of the voltages is aggravated, and the driving conditions that can be simulated during training of the neural network model are limited. At this time, it may happen that the actual driving condition is a driving condition that is not simulated when the neural network model is trained, and the SOC accuracy of the power battery obtained under the actual driving condition is low. Although higher SOC accuracy can be obtained by deepening the network complexity, this may cause problems of longer computation time and unfavorable network convergence.
The SOC of the power battery can be acquired in real time by adopting a coulomb counting method on the electric automobile, in order to correct accumulated errors of the SOC, data such as voltage, current and temperature collected by a sensor can be transmitted to a data platform, then the SOC of the battery is determined based on a data driving method, and the SOC determined on the data platform and the SOC determined on the electric automobile are compared and corrected. Due to the limited data allowed on the transmission link and the limited storage space of the data platform in practical situations, the GBT 32960.3-2016 technical specification part 3 of the communication protocol and data format of the electric vehicle remote service and management system clearly shows that the maximum time period allowed for reporting data should not exceed 30s during the normal driving of the electric vehicle. Currently, the time interval of data on a data platform is usually between 10s and 30 s. In addition, because the fluctuation frequency of the battery data is high and does not meet the sampling theorem, the data on the data platform often has a large time interval and is in an under-sampling condition, the detailed information of the battery data is reduced, and the accuracy of estimating the SOC of the power battery by using the data-driven method is reduced. Hereinafter, the "power battery" will be simply referred to as "battery".
In order to solve the above problems, an embodiment of the present application provides a method for determining a state of charge of a power battery based on data driving, which includes acquiring raw data of a battery, such as voltage, current, and temperature, through a sensor, processing the raw data, and extracting features reflecting a change trend of the raw data, features reflecting detailed information, and the like from the raw data to reduce randomness and volatility of the raw data; and then the SOC of the battery is determined according to the extracted features, and the accuracy of determining the SOC of the battery is improved. In addition, the extracted features can be corrected, the randomness and the fluctuation of original data of the electric automobile under different driving conditions are further reduced, and the accuracy of determining the SOC of the battery is improved.
The method provided by the embodiment of the application can be used for the under-sampling condition and the non-under-sampling condition. Taking the under-sampling condition as an example, a method 100 for determining the state of charge of the power battery based on data driving provided by the embodiment of the present application is described with reference to fig. 1. As shown in fig. 1, the method 100 includes:
s101: first data of the battery is acquired, and the first data comprises a plurality of voltage data of the battery.
S102: and processing the first data to obtain second data, wherein the second data comprises first characteristics, and the first characteristics reflect the change trend of the plurality of voltage data.
S103: and determining the state of charge of the battery according to the second data.
First, first data of the battery is obtained from the data platform, the first data includes a plurality of voltage data of the battery, and in addition, the first data may also include other data, such as a plurality of current data and a plurality of temperature data, that is, the composition of the first data may be (voltage, current, temperature), wherein the plurality of data are data corresponding to a plurality of moments in time within a period of time, and the data may be obtained by using a time sequence XkIn this case, k represents time. The data platform stores the raw data of the voltage, the current, the temperature and the like of the battery acquired by the sensor.
Then, in this embodiment of the application, the first data may be processed by using methods such as wavelet transform or Empirical Mode Decomposition (EMD) to obtain the second data.
Empirical mode decomposition is used as an example for explanation.
Given a time series Xk(k ═ 1,2, …, N), the process of EMD processing is as follows:
identifying a time series XkSequentially obtaining the time sequence X by a cubic spline interpolation functionkUpper envelope u of (1)kAnd the lower envelope lk. Calculating the average envelope ml of the upper and lower envelopes by equation (1)k。
Figure BDA0003252832670000081
From time series XkMinus the mean envelope mlkObtaining a new time series hlkAs shown in formula (2).
hlk＝Xk-mlk (2)
③ judging the time sequence hlkWhether two characteristics of the eigenmode functions (IMFs) are satisfied. One characteristic is: the difference between the number of the local extreme points and the number of the zero-crossing points is at most 1; another characteristic is: the average of the upper and lower envelope lines formed by the local extrema is 0. If the time sequence hlkSatisfying the above two characteristics, the time sequence hlkDefined as an IMF, and using the residue term r calculated by equation (3)kInstead of the time series Xk. And then repeating the step one to the step three.
rk＝Xk-hlk (3)
If the time sequence hlkNot satisfying both of the above characteristics of IMFs, in time series hlkOn the basis of the method, the first step to the second step are executed, and then the third step is further executed.
The above process is repeated until the termination condition of the iterative process is satisfied. The termination condition is determined according to the Cauchy convergence criterion, as shown in equation (4).
Figure BDA0003252832670000091
Wherein r isk,lastAnd rk,currentThe residual terms of the last iteration and the current iteration, respectively. In the embodiment of the present application, value may be set to 0.2.
After the iteration is finished, the residual component R can be obtainedk，RkReflects the time sequence XkCan also obtain the variation trend ofTo time series XkM IMFs, which characterize the time series XkIntrinsic vibration modes. Original time series XkMay pass IMFs and residual component RkThe reconstruction is performed as shown in formula (5).
Figure BDA0003252832670000092
When time series XkFor voltage data, the IMFs and the residual component of the voltage may be obtained, the residual component of the voltage corresponding to the first characteristic. The IMFs of the voltages are high frequency components of the plurality of voltage data, corresponding to the second characteristic. When time series XkFor current data, the IMFs and residual components of the current may be obtained, the residual components of the current corresponding to the third characteristic. The IMFs of the current are high-frequency components of the plurality of current data, and correspond to the fourth feature.
When the first data includes a plurality of voltage data, the second data includes a residual component of the voltage and may further include IMFs of the voltage. When the first data further includes a plurality of current data, the second data may further include residual components of the current and IMFs of the current. In addition, when the first data further includes a plurality of temperature data, the second data further includes temperature data.
And then inputting the second data into a pre-trained neural network model, namely determining the SOC of the battery.
However, under different working conditions, certain deviation still exists between the voltage residual components obtained by EMD processing corresponding to the same SOC, and the deviation is caused by different distribution conditions of current data under different working conditions.
The compensation strategy is specifically described below.
In consideration of the fact that under-sampling condition, system deviation can be introduced into a complex model in the discretization process, the power battery is modeled by the internal resistance model, and a corresponding measurement equation of the model is shown in a formula (6).
Uk＝OCVk+IkRinternal+αk (6)
Wherein, UkIs the voltage at time k, IkCurrent at time k, OCVkIs the open circuit voltage at time k, RinternalIs the internal resistance of the battery, alphakIs an error term at time k, and follows a standard normal distribution.
The voltage residual component is a voltage trend quantity, reflects low-frequency information of the voltage, and can be regarded as a substrate of the voltage, and the original voltage signal reconstruction is carried out on the basis of other voltages IMFs. During the charging and discharging process, the OCV changes relatively slowly and belongs to a low-frequency component. In addition, the current distribution of the operating mode can be regarded as a fluctuation in an average current, and the average voltage on the internal resistance also belongs to the low-frequency component. Therefore, in the embodiment of the present application, the voltage residual component is equivalent to the sum of the OCV and the average voltage on the internal resistance, as shown in equation (7).
Figure BDA0003252832670000101
Wherein R isk,vIs the residual component of the voltage at time k,
Figure BDA0003252832670000102
is the average current over a period of time, betakIs an error term at time k, and follows a standard normal distribution.
The linear regression equation for the internal resistance parameter identification can be obtained by subtracting the expression (7) from the expression (6), as shown in the expression (8).
Figure BDA0003252832670000103
Wherein, Deltak＝Uk-Rk,v，γkIs an error term at time k, and follows a standard normal distribution.
The parameter identification algorithm can be a least square method, and the optimal estimation of the internal resistance is obtained on line through a formula (9) by combining a formula (8) and current data acquired by an actual sensor.
Figure BDA0003252832670000104
Wherein the content of the first and second substances,
Figure BDA0003252832670000105
Y＝[Δ1,Δ2,…,ΔN]T，
Figure BDA0003252832670000106
is a current I1To INIs determined by the average value of (a) of (b),
Figure BDA0003252832670000107
is the optimal estimation result of the internal resistance of the battery.
The voltage residual component is compensated by the average voltage on the internal resistance, and the OCV-like quantity is obtained, as shown in equation (10).
Figure BDA0003252832670000108
Wherein R isk,v,cIs the voltage residual component after compensation at time k.
The residual component of the voltage is compensated, namely the second data is corrected, so that the aggregation degree of different working conditions is further improved, namely the similarity of corresponding data under different working conditions and the same SOC is improved, namely the difference of battery data of the electric automobile under different operating conditions is reduced, the randomness and the volatility of the battery data are reduced, and meanwhile the correlation between the data and the SOC is ensured.
At this time, the modified second data includes the compensated voltage residual component, and may further include voltage IMFs, current residual component, current IMFs, and temperature data. And inputting the corrected second data into the neural network model to determine the SOC of the battery, thereby further improving the accuracy of the result. The neural network can effectively simulate the dynamic characteristics of the battery through network layer stacking and nonlinear activation functions, and the method provided by the application can obtain higher SOC accuracy without deepening network complexity due to the fact that the randomness and the fluctuation of original data are reduced.
Since SOC estimation is a time sequence prediction task, in the embodiment of the present application, a Long-short term memory neural network (LSTM) that is an improved version of a Recurrent Neural Network (RNN) is used to train and predict SOC, so as to prevent the gradient disappearance and gradient explosion phenomena occurring in the back propagation process.
As shown in fig. 2, the basic structure of the LSTM unit provided in the embodiment of the present application includes an input gate i, a forgetting gate f, a control gate c, and an output gate o. Equations (11) through (16) describe the workflow of the LSTM unit.
ik＝sigm(Wx,ixk+Wh,ihk-1+bi) (11)
fk＝sigm(Wx,fxk+Wh,fhk-1+bf) (12)
ok＝sigm(Wx,oxk+Wh,ohk-1+bo) (13)
ck＝tanh(Wx,cxk+Wh,chk-1+bc) (14)
Ck＝fk⊙Ck-1+ik⊙ck (15)
hk＝ot⊙tanh(Ck) (16)
Wherein x iskIs the input of time step k (k ═ 1,2, …, N), hkIs the hidden layer output of time step k, CkIs the LSTM cell state at time step k. W and b are the weight matrix and the offset, respectively. sigmm is a sigmoid function that can convert a value to a range of 0 to 1, and tanh is a hyperbolic tangent function that can convert a value to a range of-1 to 1. Formula (11) to formula (14) The gate vector of (a) determines the information that the LSTM unit memorizes updates, forgets and outputs. Finally, the LSTM cell state and the hidden layer state are updated through equations (15) through (16).
In the embodiment of the present application, the input of the neural network model includes the compensated voltage residual component, the voltage IMFs, and may further include the current residual component, the current IMFs, and the ambient temperature. Connecting each characteristic of k time as input vector x of time step k in inputk。
Fig. 3 is a framework of SOC estimation provided in the embodiment of the present application, where N is an input sequence length, and a time corresponding to the nth data is a time step N. h is0And C0The initial values of the LSTM unit state and the hidden layer output, respectively. SOCNIs the SOC estimate at time step N. The input information of each time step is output through a hidden layer and transmitted by the state of an LSTM unit, and the past information can be fully used at present. Hidden layer output h at time step NNAnd (3) combining through a Full Connected Network (FCN), wherein the FCN outputs an SOC estimation result after passing through a sigmoid function.
In summary, the embodiment of the application provides a method for determining the state of charge of a power battery based on data driving, which performs feature extraction on original data of the battery based on EMD to obtain voltage IMFs and residual components, and IMFs and residual components of current, so that randomness and volatility of the original data are reduced.
And the internal resistance of the power battery is identified in real time by using a least square method, and the residual voltage component is compensated by the average voltage of the internal resistance, so that the data difference of the electric automobile under different operating conditions is further reduced.
The method provided by the embodiment Of the application is suitable for both under-sampling conditions and non-under-sampling conditions, and when the method is applied to a data platform, good accuracy can be ensured, the SOC on the electric automobile can be calibrated, and reference can be provided for estimating the battery capacity (SOH) according to the data Of the platform. In addition, the method can also be used for other devices with undersampling conditions, and the SOC estimation accuracy of the device is improved.
The accuracy of the methods provided in the examples of the present application is illustrated below with reference to experimental data.
And simulating undersampled battery data on a simulation data platform by using the downsampled laboratory working condition data in the experiment. The laboratory working condition data includes 9 working conditions, which are cycles 1 to 4, Neural Networks (NN), Urban power meter driving plans (UDDS), Unified driving plans (LA 92), highway fuel saving driving plans (HWFET), and high-speed driving plans (US 06). Cycle 1-4 is obtained by randomly mixing working conditions such as NN, UDDS, LA92, HWFET, US06 and the like.
In the embodiment of the application, the experimental object is a panasonic NCR18650PF ternary lithium battery, and the main performance parameters of the ternary lithium battery are shown in Table 1.
TABLE 1 NCR18650PF Performance parameters
Nominal capacity 2900mAh
Nominal voltage 3.6V
Minimum/maximum voltage 2.5V/4.2V
Standard charging current 1.35A
Maximum sustained discharge current 10A
The specific experimental procedures are as follows:
firstly, acquiring first data, and selecting a voltage, a current and a temperature time sequence by adopting a sliding window frame with the size of N.
For example, taking undersampled data with a sampling interval of 10s as an example, when N is 20 and the sliding step is 5, the data of 0 th, 10 th and 20 th seconds … …, 200 th seconds are taken as the first group of data, and the data of 50 th, 60 th and 70 th seconds … …, 250 th seconds are taken as the second group of data.
Acquiring second data, decomposing the voltage and current signals by using EMD, acquiring voltage IMFs and residual components, and acquiring current IMFs and residual components, wherein the second data also comprises temperature data.
And thirdly, establishing a linear regression equation for internal resistance parameter identification based on the formula (8), acquiring an optimal solution of the internal resistance through a formula (9) based on a least square method, and compensating the voltage residual component by adopting a formula (10).
And fourthly, dividing the 9 working condition data sets into a training set and a test set, wherein the HWFET and US06 working conditions are the test set, and other working conditions are the training set. The training set and the test set are normalized.
Setting the time step length and the hidden layer size of the LSTM model, and training the model by using a training set. In the training process, the parameter optimization process adopts an adam algorithm, the training step length is 2000, the learning rate is set to 0.01, and the batch size is set to 256.
Sixthly, comparing and analyzing the SOC estimation result of the LSTM based on the first data (hereinafter, referred to as a standard LSTM) and the SOC estimation result of the LSTM based on the second data (hereinafter, referred to as a new feature LSTM).
The experimental results were analyzed as follows.
Figure 4 is a data plot of the voltage, current and temperature of the cell at ambient temperatures-10 c and 25 c for the US06 operating conditions.
FIGS. 5(a) and (d) are graphs of the pre-and post-compensation voltage residual components for 60% SOC at-10 deg.C for Cycle 1, NN, UDDS, LA92, HWFET, and US06 operating conditions.
FIGS. 5(b) and (e) are graphs of the pre-and post-compensation voltage residual components for 50% SOC at-10 deg.C for Cycle 1, NN, UDDS, LA92, HWFET, and US06 operating conditions.
FIGS. 5(c) and (f) are graphs of the pre-and post-compensation voltage residual components for 40% SOC at-10 deg.C for Cycle 1, NN, UDDS, LA92, HWFET, and US06 operating conditions.
The standard deviation of the voltage residual component is used to measure the aggregation degree of different working condition data. Table 2 shows the standard deviation of the residual voltage components before and after compensation corresponding to 20% SOC to 80% SOC under the working conditions of environment temperature of-10 ℃, Cycle 1, NN, UDDS, LA92, HWFET and US06, and the SOC interval is 10%. stdvrAnd stdcvrRespectively, the standard deviation of the residual voltage components before and after compensation under different working conditions. As can be seen from fig. 5 and table 2, in a low-temperature environment, the standard deviation of the compensated voltage residual component is generally reduced, the aggregation degree of data under different working conditions is improved, and the effectiveness of the compensation strategy is proved.
TABLE 2 Standard deviations of residual components of voltages before and after compensation under different conditions of different SOC at ambient temperature-10 deg.C
Figure BDA0003252832670000141
FIGS. 6 and 7 are SOC estimates and error comparisons for HWFET and US06 operating conditions at ambient temperatures of-10 deg.C and 25 deg.C, respectively, using the two methods. The two methods are the new feature LSTM and the standard LSTM, respectively. Table 3 gives the estimation error statistics for both methods.
As can be seen from fig. 6, 7 and table 3, when the ambient temperature is normal temperature, the SOC estimation accuracy is better based on both the new characteristic LSTM and the standard LSTM. When the ambient temperature is low, the estimation result based on the new feature LSTM is better than that of the standard LSTM, which is caused by the new feature (second data) improving the degree of aggregation in different conditions.
It can be seen that the maximum Root Mean Square Error (RMSE) was 3.7% and the maximum absolute error (MaxAE) was 5.4% of the SOC estimates for HWFETs and US06 for 4 ambient temperature conditions based on standard LSTM; among the SOC estimates obtained for HWFET and US06 for 4 ambient temperature conditions based on the new signature LSTM, the RMSE maximum was 2.2% and the MaxAE maximum was 5.6%. The above results fully illustrate the stability of the method proposed in the examples of the present application at different ambient temperatures.
TABLE 3 SOC estimation error statistics for two methods
Figure BDA0003252832670000151
The following describes an apparatus and an electronic device provided in an embodiment of the present application.
Fig. 8 is an apparatus for training a model according to an embodiment of the present application, where the apparatus 800 includes an obtaining unit 801 and a processing unit 802.
The acquiring unit 801 is configured to acquire first data of a battery, where the first data includes a plurality of voltage data of the battery.
The processing unit 802 is configured to process the first data to obtain second data, where the second data includes a first characteristic, and the first characteristic reflects a variation trend of the plurality of voltage data; and determining the state of charge of the battery according to the second data.
In particular, the processing unit 802 is further configured to process the first data according to an empirical mode decomposition method to obtain the second data.
In particular, the processing unit 802 is further configured to obtain an internal resistance of the battery; correcting the second data according to the internal resistance of the battery to obtain corrected second data; and determining the state of charge of the battery according to the corrected second data.
In particular, the processing unit 802 is further configured to input the modified second data into a neural network model to obtain a state of charge of the battery, and the neural network model is configured to output the state of charge of the battery according to the first characteristic.
In particular, the second data further includes a second feature that is a high frequency component of the plurality of voltage data.
In particular, the first data further includes a plurality of current data and a plurality of temperature data of the battery
Specifically, the second data further includes a third feature reflecting a trend of change of the plurality of current data, a fourth feature being a high frequency component of the plurality of current data, and temperature data.
It should be understood that the apparatus 800 of the embodiment of the present application may be implemented by an application-specific integrated circuit (ASIC), or a Programmable Logic Device (PLD), which may be a Complex Programmable Logic Device (CPLD), a field-programmable gate array (FPGA), a General Array Logic (GAL), or any combination thereof. The method shown in fig. 1 may also be implemented by software, and when the method shown in fig. 1 is implemented by software, the apparatus 800 and each module thereof may also be a software module.
Fig. 9 is a schematic structural diagram of an electronic device 900 according to an embodiment of the present application. As shown in fig. 9, the device 900 includes a processor 901, a memory 902, a communication interface 903, and a bus 904. The processor 901, the memory 902, and the communication interface 903 communicate with each other via the bus 904, and may also communicate with each other by other means such as wireless transmission. The memory 902 is used for storing instructions and the processor 901 is used for executing the instructions stored by the memory 902. The memory 902 stores program code 9021, and the processor 901 may call the program code 9021 stored in the memory 902 to perform the method shown in fig. 1.
It should be understood that in the embodiments of the present application, the processor 901 may be a CPU, and the processor 901 may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general purpose processor may be a microprocessor or any conventional processor or the like.
The memory 902 may include a read-only memory and a random access memory, and provides instructions and data to the processor 901. The memory 902 may also include non-volatile random access memory. The memory 902 may be either volatile memory or nonvolatile memory, or may include both volatile and nonvolatile memory. The non-volatile memory may be a read-only memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an electrically Erasable EPROM (EEPROM), or a flash memory. Volatile memory can be Random Access Memory (RAM), which acts as external cache memory. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (static RAM, SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and direct bus RAM (DR RAM).
The bus 904 may include a power bus, a control bus, a status signal bus, and the like, in addition to a data bus. But for clarity of illustration the various busses are labeled in figure 9 as bus 904.
The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, the above-described embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded or executed on a computer, cause the processes or functions described in accordance with the embodiments of the application to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more collections of available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium may be a Solid State Drive (SSD).
The above-mentioned embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present application and are intended to be included within the scope of the present application.

Claims (10)

1. A method for determining a state of charge of a power battery based on data driving, the method comprising:
acquiring first data of a battery, wherein the first data comprises a plurality of voltage data of the battery;
processing the first data to obtain second data, wherein the second data comprises first characteristics, and the first characteristics reflect the change trend of the voltage data;
and determining the state of charge of the battery according to the second data.
2. The method of claim 1, wherein the processing the first data to obtain second data comprises:
and processing the first data according to an empirical mode decomposition method to obtain the second data.
3. The method of claim 2, wherein after obtaining the second data, the method further comprises:
obtaining the internal resistance of the battery;
correcting the second data according to the internal resistance of the battery to obtain corrected second data;
determining the state of charge of the battery according to the second data, comprising:
and determining the state of charge of the battery according to the corrected second data.
4. The method of claim 3, wherein said determining the state of charge of the battery from said modified second data comprises:
and inputting the corrected second data into a neural network model to obtain the state of charge of the battery, wherein the neural network model is used for outputting the state of charge of the battery according to the first characteristic.
5. The method of any of claims 1 to 4, wherein the second data further comprises a second characteristic, the second characteristic being a high frequency component of the plurality of voltage data.
6. The method of any of claims 1 to 4, wherein the first data further comprises a plurality of current data and a plurality of temperature data for the battery.
7. The method of claim 6, wherein the second data further comprises a third characteristic reflecting a trend of the plurality of current data, a fourth characteristic being a high frequency component of the plurality of current data, and temperature data.
8. An apparatus for determining a state of charge of a power battery based on data driving, the apparatus comprising:
an acquisition unit configured to acquire first data of a battery, the first data including a plurality of voltage data of the battery;
the processing unit is used for processing the first data to obtain second data, wherein the second data comprises first characteristics, and the first characteristics reflect the change trend of the plurality of voltage data; and determining the state of charge of the battery according to the second data.
9. An electronic device, comprising: a memory storing a computer program and a processor implementing the method of any one of claims 1 to 7 when the processor executes the computer program.
10. A computer-readable storage medium having stored thereon computer instructions which, when run on an electronic device, cause the electronic device to perform the method of any one of claims 1-7.
CN202111051231.1A 2021-09-08 2021-09-08 Method for determining state of charge of power battery based on data driving Active CN113933725B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111051231.1A CN113933725B (en) 2021-09-08 2021-09-08 Method for determining state of charge of power battery based on data driving

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111051231.1A CN113933725B (en) 2021-09-08 2021-09-08 Method for determining state of charge of power battery based on data driving

Publications (2)

Publication Number Publication Date
CN113933725A true CN113933725A (en) 2022-01-14
CN113933725B CN113933725B (en) 2023-09-12

Family

ID=79275349

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111051231.1A Active CN113933725B (en) 2021-09-08 2021-09-08 Method for determining state of charge of power battery based on data driving

Country Status (1)

Country Link
CN (1) CN113933725B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114579644A (en) * 2022-05-06 2022-06-03 中国汽车技术研究中心有限公司 Method, apparatus and medium for battery efficient charging data identification based on deep learning

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150081237A1 (en) * 2013-09-19 2015-03-19 Seeo, Inc Data driven/physical hybrid model for soc determination in lithium batteries
CN106154168A (en) * 2016-04-01 2016-11-23 储盈新能源科技（上海）有限公司 The method for estimating charge state of power cell of data-driven
US20180143257A1 (en) * 2016-11-21 2018-05-24 Battelle Energy Alliance, Llc Systems and methods for estimation and prediction of battery health and performance
EP3410139A1 (en) * 2017-05-29 2018-12-05 Magneti Marelli S.p.A. Method for estimating the current and the state of charge of a battery pack or cell, without direct detection of current under operating conditions
CN110941929A (en) * 2019-12-06 2020-03-31 长沙理工大学 Battery health state assessment method based on ARMA and Elman neural network combined modeling
CN111220921A (en) * 2020-01-08 2020-06-02 重庆邮电大学 Lithium battery capacity estimation method based on improved convolution-long-and-short-term memory neural network
CN112630662A (en) * 2020-12-30 2021-04-09 南京航空航天大学 Power battery SOH estimation method based on data driving and multi-parameter fusion
CN112782594A (en) * 2020-12-24 2021-05-11 杭州电子科技大学 Method for estimating SOC (state of charge) of lithium battery by data-driven algorithm considering internal resistance
CN113033104A (en) * 2021-03-31 2021-06-25 浙江大学 Lithium battery state of charge estimation method based on graph convolution
CN113075554A (en) * 2021-03-26 2021-07-06 国网浙江省电力有限公司电力科学研究院 Lithium ion battery pack inconsistency identification method based on operation data
US20210256901A1 (en) * 2020-02-18 2021-08-19 InveStar Holdings Inc. System and method for modulating an array of emissive elements

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150081237A1 (en) * 2013-09-19 2015-03-19 Seeo, Inc Data driven/physical hybrid model for soc determination in lithium batteries
CN106154168A (en) * 2016-04-01 2016-11-23 储盈新能源科技（上海）有限公司 The method for estimating charge state of power cell of data-driven
US20180143257A1 (en) * 2016-11-21 2018-05-24 Battelle Energy Alliance, Llc Systems and methods for estimation and prediction of battery health and performance
EP3410139A1 (en) * 2017-05-29 2018-12-05 Magneti Marelli S.p.A. Method for estimating the current and the state of charge of a battery pack or cell, without direct detection of current under operating conditions
CN110941929A (en) * 2019-12-06 2020-03-31 长沙理工大学 Battery health state assessment method based on ARMA and Elman neural network combined modeling
CN111220921A (en) * 2020-01-08 2020-06-02 重庆邮电大学 Lithium battery capacity estimation method based on improved convolution-long-and-short-term memory neural network
US20210256901A1 (en) * 2020-02-18 2021-08-19 InveStar Holdings Inc. System and method for modulating an array of emissive elements
CN112782594A (en) * 2020-12-24 2021-05-11 杭州电子科技大学 Method for estimating SOC (state of charge) of lithium battery by data-driven algorithm considering internal resistance
CN112630662A (en) * 2020-12-30 2021-04-09 南京航空航天大学 Power battery SOH estimation method based on data driving and multi-parameter fusion
CN113075554A (en) * 2021-03-26 2021-07-06 国网浙江省电力有限公司电力科学研究院 Lithium ion battery pack inconsistency identification method based on operation data
CN113033104A (en) * 2021-03-31 2021-06-25 浙江大学 Lithium battery state of charge estimation method based on graph convolution

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
FANG LIU ET AL.: "SOC estimation based on data driven exteaded Kalman filter algorithm for power battery of electric vehicle and plug-in electric vehicle", JOURNAL OF CENTRAL SOUTH UNIVERSITY, vol. 2018, no. 6, pages 1402 - 1415, XP036832770, DOI: 10.1007/s11771-019-4096-5 *
王坚等: "井筒变形监测理论与方法", vol. 2018, 武汉大学出版社, pages: 135 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114579644A (en) * 2022-05-06 2022-06-03 中国汽车技术研究中心有限公司 Method, apparatus and medium for battery efficient charging data identification based on deep learning

Also Published As

Publication number Publication date
CN113933725B (en) 2023-09-12

Similar Documents

Publication Publication Date Title
CN112540317B (en) Battery health state estimation and residual life prediction method based on real vehicle data
Zhao et al. Elman neural network using ant colony optimization algorithm for estimating of state of charge of lithium-ion battery
CN112946499A (en) Lithium battery health state and charge state joint estimation method based on machine learning
CN113156321B (en) Estimation method of lithium ion battery state of charge (SOC)
CN111426957B (en) SOC estimation optimization method for power battery under simulated vehicle working condition
Sun et al. A sequential capacity estimation for the lithium-ion batteries combining incremental capacity curve and discrete Arrhenius fading model
WO2023184824A1 (en) Method and system for estimating state of charge of battery cluster, electronic device, and storage medium
CN112816879A (en) Online estimation method for power battery SoE for variable-working-condition vehicle
CN114371408B (en) Method for estimating state of charge of battery, method and device for extracting charging curve
Jin et al. SOC estimation of lithium-ion battery considering the influence of discharge rate
CN114290960A (en) Method and device for acquiring battery health degree of power battery and vehicle
Dong et al. Low complexity state-of-charge estimation for lithium-ion battery pack considering cell inconsistency
CN114594378A (en) Lithium battery SOH model training method, prediction method, device and equipment
CN113933725B (en) Method for determining state of charge of power battery based on data driving
CN117665620A (en) New energy automobile data-based battery health evaluation method
CN116400247B (en) Method and device for determining soft short circuit fault of battery
CN116593917A (en) Lithium battery health state estimation method based on time-frequency double-flow characteristic characterization
CN116699414A (en) Lithium battery SOC estimation method and system based on UKF-LSTM algorithm
CN115856641A (en) Method and device for predicting remaining charging time of battery and electronic equipment
CN116400223A (en) Battery electric quantity information generation method, device and readable storage medium
CN113447826B (en) SOC determination method and device based on steady-state equivalent circuit model
CN112578286B (en) Battery SOC estimation method and device
CN115219907A (en) Lithium battery SOC estimation method, system, medium, equipment and terminal
CN114487850A (en) Power battery capacity prediction method based on real vehicle data
CN114295987A (en) Battery SOC state estimation method based on nonlinear Kalman filtering

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant