CN114814618A - Lithium ion battery residual capacity estimation method, device, equipment and storage medium - Google Patents

Lithium ion battery residual capacity estimation method, device, equipment and storage medium Download PDF

Info

Publication number
CN114814618A
CN114814618A CN202210604284.XA CN202210604284A CN114814618A CN 114814618 A CN114814618 A CN 114814618A CN 202210604284 A CN202210604284 A CN 202210604284A CN 114814618 A CN114814618 A CN 114814618A
Authority
CN
China
Prior art keywords
capacity
lithium ion
voltage
increment
battery capacity
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
Application number
CN202210604284.XA
Other languages
Chinese (zh)
Inventor
张梦帆
冯华
钟伟
尹启睿
谭志阳
刘子璇
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.)
Shanghai Thinktech Information Technology Co ltd
Original Assignee
Shanghai Thinktech Information Technology Co ltd
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 Shanghai Thinktech Information Technology Co ltd filed Critical Shanghai Thinktech Information Technology Co ltd
Priority to CN202210604284.XA priority Critical patent/CN114814618A/en
Publication of CN114814618A publication Critical patent/CN114814618A/en
Pending legal-status Critical Current

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
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/392—Determining battery ageing or deterioration, e.g. state of health
    • 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]
    • G01R31/396—Acquisition or processing of data for testing or for monitoring individual cells or groups of cells within a battery
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Abstract

The application relates to a method, a device, equipment and a storage medium for estimating the residual capacity of a lithium ion battery in the technical field of lithium ion batteries. The method comprises the following steps: the method comprises the steps of obtaining experimental measurement values of a plurality of lithium ion batteries, extracting capacity increment wave peaks from results obtained by fitting the measurement values in an SVM mode, conducting preliminary capacity prediction by respectively adopting an SVM, an LSTM network and a GRP, then taking the preliminary capacity prediction as input, and conducting fusion by utilizing a random forest algorithm to obtain capacity output. The method combines the advantages of the SVM in the nonlinear and high-dimensional space fitting problem, the GRP in the uncertainty prediction problem and the LSTM network in the time sequence prediction problem, effectively solves the interference of voltage acquisition noise on the capacity increment curve, and solves the difficulty in effectively extracting the characteristics of the capacity increment curve. Meanwhile, the random forest algorithm is adopted to estimate the battery capacity, and the defect that a single machine learning algorithm is easy to fall into local optimization is overcome.

Description

Lithium ion battery residual capacity estimation method, device, equipment and storage medium
Technical Field
The present disclosure relates to the field of lithium ion battery technologies, and in particular, to a method, an apparatus, a device, and a storage medium for estimating a remaining capacity of a lithium ion battery.
Background
The lithium ion battery is a clean and efficient energy storage system and is widely popularized in electric automobiles. However, the capacity of the lithium ion battery inevitably gradually deteriorates due to the material characteristics thereof. Therefore, accurate monitoring of the remaining capacity of the lithium ion battery not only facilitates estimation and promotion of the state of health and the state of charge, but also plays a crucial role in ensuring reliable operation of the electric vehicle.
As computer science has developed, machine learning methods are gradually introduced into the capacity estimation problem. In a machine learning approach, capacity is estimated by mapping potential non-linear relationships between healthy features and remaining capacity. The capacity Increment (IC) curve is an important health feature that has been shown to have a strong correlation with battery capacity, and it can reconstruct the voltage plateau caused by the internal electrochemical reaction equilibrium as an intuitive and recognizable peak or trough. However, the IC curve is sensitive to noise, and when the measured voltage is noisy, the monotonicity of the voltage cannot be guaranteed, so that the denominator of the IC calculation is zero or even negative, which presents a great challenge to the calculation of the IC curve.
In the existing machine learning method, a long-term memory recurrent neural network is a typical neural network algorithm, and long-term dependence information is captured through a specially designed gate structure. Compared with the traditional neural network, the long-time memory recurrent neural network has higher nonlinear modeling capability and more accurate estimation performance when processing time series data. The Gaussian process regression performs prediction distribution estimation on the parameters of a specific time point instead of single deterministic point estimation, and is beneficial to uncertainty quantitative estimation. However, this method suffers from a decrease in estimation accuracy when a high-dimensional space is encountered. The support vector machine algorithm is a famous classification regression method, and shows high efficiency in solving the fitting problem of nonlinear and high-dimensional models. Although these methods can achieve capacity estimation, they are all based on a single learner, which tends to trap the estimation result into local optima, thus limiting the fidelity of the capacity estimation.
Disclosure of Invention
In view of the above, it is necessary to provide a method, an apparatus, a device and a storage medium for estimating remaining capacity of a lithium ion battery.
A lithium ion battery remaining capacity estimation method, the method comprising:
and acquiring multiple groups of measured values of a plurality of target lithium ion batteries in an aging experiment, wherein the measured values comprise the voltage, the current and the capacity of the target lithium ion batteries.
And obtaining capacity increment data according to the voltage capacity relation obtained by fitting by adopting a fitting mode of a support vector machine according to the measured value, and extracting a capacity increment peak of the capacity increment data.
And taking a plurality of capacity increment wave peaks as training samples.
And respectively inputting the training samples into the trained SVM, LSTM network and Gaussian process regression operator to respectively obtain a first predicted battery capacity, a second predicted battery capacity and a third predicted battery capacity.
And inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into a trained random forest operator to obtain a target lithium ion battery capacity.
In one embodiment, acquiring multiple sets of measurement values of a target lithium ion battery in an aging experiment includes:
and (3) carrying out experiments on a plurality of target lithium ion batteries under a preset test environment, and carrying out aging experiments by repeated charge and discharge.
In the cycle life test, after a preset aging period, a preset rated capacity value is subjected to charge-discharge cycle, the current capacity is calibrated, and data acquisition is carried out by adopting a preset sampling frequency, so that the voltage, the current and the capacity of the target lithium ion battery are obtained.
In one embodiment, obtaining capacity increment data according to a voltage capacity relationship obtained by fitting by using a fitting method of a support vector machine according to the measured value, and extracting a capacity increment peak of the capacity increment data includes:
and taking the voltage of the measured value as input and the capacity as output, and obtaining a voltage-capacity relation by adopting a fitting mode of a support vector machine.
Calculating the voltage-capacity relation by adopting a capacity increment data calculation formula to obtain capacity increment data; the capacity increment data calculation formula is as follows:
Figure BDA0003670721480000031
wherein:
Figure BDA0003670721480000032
which represents the differential of the capacity versus the voltage,
Figure BDA0003670721480000033
representing the ratio of capacity increment to voltage increment, Q t 、Q t-1 Respectively representing the capacity values, V, at time t and t-1 t 、V t-1 Respectively representing the voltage values at time t and t-1.
And extracting a capacity increment peak in the capacity increment data, and taking the capacity increment peak as a health characteristic parameter.
In one embodiment, the inputting the training samples into the trained SVM, LSTM network, and gaussian process regression operator respectively to obtain a first predicted battery capacity, a second predicted battery capacity, and a third predicted battery capacity includes:
and inputting the training samples into a trained SVM, characterizing the training samples into a high-dimensional space characteristic by adopting a kernel function, and obtaining a first predicted battery capacity by adopting the high-dimensional space characteristic by adopting a nonlinear SVM.
And inputting the training sample into a trained LSTM network, and capturing the long-term dependence relationship through a gate structure to obtain the second battery capacity.
And inputting the training sample into a trained Gaussian process regression operator to obtain the third battery capacity.
In one embodiment, inputting the first predicted battery capacity, the second predicted battery capacity, and the third predicted battery capacity into a trained random forest operator to obtain a target lithium ion battery capacity, includes:
inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into a trained random forest operator to obtain the output of each tree in the random forest; the output expression of the jth tree in the forest of the random forest operator is as follows:
Figure BDA0003670721480000034
wherein, c j Output of jth tree in forest, X, representing random forest operator i It is indicated that the point (i) is,
Figure BDA0003670721480000041
represents a training sample set, Θ j It is meant that the random variables are distributed the same,
Figure BDA0003670721480000042
representing all of the cells containing x,
Figure BDA0003670721480000043
indicates all fall in
Figure BDA0003670721480000044
The number of points.
Adopting a bagging strategy according to the output of each tree in the random forest to obtain the target lithium ion battery capacity; the expression of the bagging strategy of the random forest operator is as follows:
Figure BDA0003670721480000045
wherein the content of the first and second substances,
Figure BDA0003670721480000046
the results of the random forest estimation are shown,
Figure BDA0003670721480000047
the estimation result of the jth tree in the random forest is shown, and M represents the number of trees.
A lithium ion battery remaining capacity estimation apparatus, the apparatus comprising:
the test data acquisition module is used for acquiring a plurality of groups of measured values of a plurality of target lithium ion batteries in an aging test, wherein the measured values comprise voltage, current and capacity of the target lithium ion batteries.
The training sample determining module is used for obtaining capacity increment data according to the voltage capacity relation obtained by fitting by adopting a fitting mode of a support vector machine according to the measured value and extracting a capacity increment peak of the capacity increment data; and taking a plurality of capacity increment wave peaks as training samples.
The target lithium ion battery capacity determining module is used for respectively inputting the training samples into the trained SVM, LSTM network and Gaussian process regression operator to respectively obtain a first predicted battery capacity, a second predicted battery capacity and a third predicted battery capacity; and inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into a trained random forest operator to obtain a target lithium ion battery capacity.
In one embodiment, the experimental data acquisition module is further configured to perform an experiment on a plurality of target lithium ion batteries in a predetermined test environment, and perform an aging experiment by repeatedly charging and discharging; in a cycle life test, after a preset aging period, a preset rated capacity value is subjected to charge-discharge cycle, the current capacity is calibrated, and data acquisition is carried out by adopting a preset sampling frequency, so that the voltage, the current and the capacity of the target lithium ion battery are obtained.
In one embodiment, the training sample determining module is further configured to obtain a voltage-to-capacity relationship by using the voltage of the measured value as an input and the capacity as an output and using a fitting mode of a support vector machine; calculating the voltage-capacity relation by adopting a capacity increment data calculation formula to obtain capacity increment data; the capacity increment data calculation formula is as follows:
Figure BDA0003670721480000051
wherein:
Figure BDA0003670721480000052
which represents the differential of the capacity versus the voltage,
Figure BDA0003670721480000053
representing the ratio of capacity increment to voltage increment, Q t 、Q t-1 Respectively representing the capacity values, V, at time t and t-1 t 、V t-1 Respectively representing the voltage values at the time t and t-1;
and extracting a capacity increment peak in the capacity increment data, and taking the capacity increment peak as a health characteristic parameter.
A computer device comprising a memory storing a computer program and a processor implementing the steps of any of the methods described above when the processor executes the computer program.
A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of any of the methods described above.
The method comprises the steps of obtaining experimental measured values of a plurality of lithium ion batteries, fitting by adopting a support vector machine mode, extracting capacity increment wave peaks from fitting results, performing preliminary capacity prediction by respectively adopting a support vector machine, a long-time memory recurrent neural network and a Gaussian process regression, then taking the preliminary capacity prediction as input, and fusing the output of a multi-machine learning machine by utilizing a random forest algorithm. The method combines the advantages of a support vector machine in the nonlinear and high-dimensional space fitting problem, the Gaussian process regression in the uncertainty prediction problem and the long and short time memory of the recurrent neural network in the time sequence prediction problem, effectively solves the interference of voltage acquisition noise on the capacity increment curve, and solves the difficulty of effectively extracting the characteristics of the capacity increment curve. Meanwhile, a random forest algorithm is adopted to integrate a support vector machine, a long-time and short-time memory network and a Gaussian process regression algorithm to estimate the battery capacity, and the defect that a single machine learning algorithm is easy to fall into local optimization is effectively overcome.
Drawings
FIG. 1 is a schematic flow chart of a method for estimating remaining capacity of a lithium ion battery according to an embodiment;
FIG. 2 is a plot of charge voltage versus charge capacity for one embodiment;
FIG. 3 is a plot of a capacity delta peak versus battery capacity for one embodiment;
FIG. 4 shows an embodiment of a battery capacity estimation result;
fig. 5 is a block diagram showing a configuration of a lithium ion battery remaining capacity estimating apparatus according to an embodiment;
FIG. 6 is a diagram illustrating an internal structure of a computer device according to an embodiment.
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.
Support Vector Machine, short for: and (3) SVM.
Gaussian Process regression, Gaussian Process Regressor, for short: GPR is carried out.
A Long short-term memory recurrent neural network is called LSTM RNN for short.
In one embodiment, as shown in fig. 1, there is provided a lithium ion battery remaining capacity estimation method, including the steps of:
step 100: and acquiring multiple groups of measured values of the target lithium ion batteries in an aging experiment, wherein the measured values comprise the voltage, the current and the capacity of the target lithium ion batteries.
Specifically, charging voltage, current, and capacity data were collected from experimental tests.
Step 102: obtaining capacity increment data according to the voltage capacity relation obtained by fitting by adopting a fitting mode of a support vector machine according to the measured value, and extracting a capacity increment peak of the capacity increment data; and taking a plurality of capacity increment wave peaks as training samples.
Specifically, the charging voltage data is used as input, the charging capacity data is used as output, and the relation between the charging voltage and the charging capacity is solved by adopting a fitting mode of a support vector machine.
Extracting a capacity increment peak in the capacity increment data as a health characteristic parameter to obtain a relation between the capacity increment peak and the battery capacity, wherein the peak value of a capacity increment curve continuously decreases along with the decrease of the battery capacity
Step 104: and respectively inputting the training samples into the trained SVM, LSTM network and Gaussian process regression operator to respectively obtain a first predicted battery capacity, a second predicted battery capacity and a third predicted battery capacity.
Specifically, 1) compared with the traditional neural network, the SVM converges to the optimal value more quickly, and fitting of nonlinear and high-dimensional data can be satisfied. For the non-linear case, the raw data is characterized in SVM as a high dimensional space using a kernel function. The kernel function used in the SVM may be a linear kernel function, a gaussian radial basis kernel function, or a polynomial kernel function.
2) The LSTM network is an extension form of a traditional neural network, can capture long-term dependence through a gate structure, has higher nonlinear building potential, and has more accurate performance when time series prediction is processed. The LSTM consists of an input gate, a forgetting gate, an output gate and the different input-output connections controlled by these gates.
3) GPR mainly describes a regression function based on probability distribution, empirical risks are reflected through a likelihood function, and posterior probability distribution is obtained through a Bayes theory.
The fusion framework provided by the method combines the advantages of SVM in the nonlinear and high-dimensional space fitting problem, GPR in the uncertainty prediction problem and LSTM network in the time sequence prediction problem.
Step 106: and inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into the trained random forest operator to obtain the target lithium ion battery capacity.
Specifically, the estimation results of the SVM, LSTM network and Gaussian process regression operator are as follows: the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity are used as input of a fusion method, and output of the multi-machine learning machine is fused by using a Random Forest (RF) algorithm. RF is a typical ensemble learning method, and can be integrated with a single learner to obtain the optimal estimation result.
The method for estimating the residual capacity of the lithium ion battery comprises the steps of obtaining experimental measured values of a plurality of lithium ion batteries, fitting by adopting a support vector machine mode, extracting capacity increment wave peaks from fitting results, performing preliminary capacity prediction by respectively adopting a support vector machine, a long-time memory recurrent neural network and a Gaussian process regression, taking the preliminary capacity prediction as input, and fusing the output of a multi-machine learning machine by utilizing a random forest algorithm. The method combines the advantages of a support vector machine in the nonlinear and high-dimensional space fitting problem, the Gaussian process regression in the uncertainty prediction problem and the long and short time memory of the recurrent neural network in the time sequence prediction problem, effectively solves the interference of voltage acquisition noise on the capacity increment curve, and solves the difficulty of effectively extracting the characteristics of the capacity increment curve. Meanwhile, a random forest algorithm is adopted to integrate a support vector machine, a long-time and short-time memory network and a Gaussian process regression algorithm to estimate the battery capacity, and the defect that a single machine learning algorithm is easy to fall into local optimization is effectively overcome.
In one embodiment, step 100 comprises: carrying out experiments on a plurality of target lithium ion batteries under a preset test environment, and carrying out aging experiments by repeated charge and discharge; in the cycle life test, after a preset aging period, a preset rated capacity value is subjected to charge-discharge cycle, the current capacity is calibrated, and data acquisition is carried out by adopting a preset sampling frequency, so that the voltage, the current and the capacity of the target lithium ion battery are obtained.
Specifically, all the batteries were tested in a temperature controlled oven at 40 ℃ and repeatedly charged and discharged to show their aging characteristics. In the charging process, a constant current charging scheme of 2C (wherein C is a rated capacity value) is adopted until the terminal voltage reaches 4.2V; the battery is then charged with a constant voltage charging strategy until the current drops to a predefined cutoff threshold. In the discharging stage, the aging process in practical application is simulated by adopting a load working condition based on Artemis. In the cycle life test, 1C charge-discharge cycle is carried out every 100 aging periods, the current capacity is calibrated, the voltage, the current and the capacity of the target battery are recorded, and the sampling frequency is 1 Hz.
In one embodiment, step 102 comprises: taking the voltage of the measured value as input and the capacity as output, and obtaining a voltage-capacity relation by adopting a fitting mode of a support vector machine; calculating the voltage-capacity relation by adopting a capacity increment data calculation formula to obtain capacity increment data; the capacity increment data calculation formula is as follows:
Figure BDA0003670721480000081
wherein:
Figure BDA0003670721480000082
which represents the differential of the capacity versus the voltage,
Figure BDA0003670721480000083
representing the ratio of capacity increment to voltage increment, Q t 、Q t-1 Respectively representing the capacity values, V, at time t and t-1 t 、V t-1 Respectively representing the voltage values at the time t and t-1;
and extracting a capacity increment peak in the capacity increment data, and taking the capacity increment peak as a health characteristic parameter.
Specifically, charging voltage data is used as input, charging capacity data is used as output, an SVM algorithm is adopted to fit the relationship between the charging voltage and the charging capacity, the fitting result is shown in the following figure 2, and as can be seen from figure 2, the fitted curve of the SVM eliminates the influence of noise, and the relationship between the voltage and the capacity is well fitted.
And (4) processing the fitting data by using the formula (1) to obtain capacity increment data. The capacity increment peak in the capacity increment data is extracted as a health characteristic parameter, and the relationship between the capacity increment peak and the battery capacity shown in fig. 3 is obtained, and as can be seen from fig. 3, the peak value of the capacity increment curve is continuously decreased along with the decrease of the battery capacity. Therefore, this parameter is applied as a health feature mining to the estimation of the battery capacity.
In one embodiment, step 104 includes: inputting a training sample into a trained SVM, characterizing the training sample into a high-dimensional space feature by adopting a kernel function, and obtaining a first predicted battery capacity by adopting a nonlinear SVM for the high-dimensional space feature; inputting the training sample into a trained LSTM network, and capturing the long-term dependence relationship through a gate structure to obtain a second battery capacity; and inputting the training sample into a trained Gaussian process regression operator to obtain the third battery capacity.
Specifically, 1) compared with the traditional neural network, the SVM converges to the optimal value more quickly, and fitting of nonlinear and high-dimensional data can be satisfied. For the non-linear case, the raw data is characterized in the SVM as a high dimensional space using a kernel function. The nonlinear problem of the SVM is derived as follows:
Figure BDA0003670721480000091
in the formula, epsilon represents tolerance deviation, omega and b represent weight and deviation, and lagrange multipliers alpha and alpha can be introduced * To solve, namely:
Figure BDA0003670721480000092
2) the LSTM network is an extension form of a traditional neural network, can capture long-term dependence through a gate structure, has higher nonlinear building potential, and has more accurate performance when time series prediction is processed. The LSTM network consists of an input gate, a forgetting gate, an output gate and the different input and output connections controlled by these gates. The calculation process of the LSTM network is summarized as follows:
Figure BDA0003670721480000093
wherein f is k ，i k ，g k O and c k Respectively representing a forgetting gate, an input node, an output gate and a storage unit; b is a mixture of f ，b i ，b g Respectively representing the deviations of the forgetting gate, the input gate and the input node, OP k Indicating the output, OW, at a representation step k f 、OW i 、OW g 、OW O Output weights, IW, for the forgetting gate, the input node and the output gate, respectively f 、IW i 、IW g 、IW O Input weights of the forgetting gate, the input node and the output gate are respectively; IP (Internet protocol) k And IP k-1 Corresponding to the inputs at representing step k and k-1; p is a radical of k Internal variables representing LSTM storage locations; sigma is an activation function, and preferably, the activation function adopts a sigmoid function; tanh is defined as a hyperbolic function.
3) GPR mainly describes a regression function based on probability distribution, empirical risks are reflected through a likelihood function, and posterior probability distribution is obtained through a Bayes theory. Therefore, the GPR problem can be simplified to:
Figure BDA0003670721480000101
where f (x) is the output function,
Figure BDA0003670721480000102
() Gaussian probability distribution function, x and y representing input variables and observed variables, m (x) and k f (x, x') is the mean and covariance functions, ζ represents the added noise,
Figure BDA0003670721480000103
y、y * and
Figure BDA0003670721480000104
respectively representing prior value, predicted value, mean value of predicted value, x and x * N input vectors representing high dimensions and test input vector
Figure BDA0003670721480000105
Representing the noise covariance matrix, p (y) * ∣x,y,x * ) Represents a prior distribution, K f (x, x) denotes an n-dimensional symmetric positive definite matrix.
The training processes of the SVM, the LSTM network, the Gaussian process regression and the random forest are similar, the peak value of a 70% capacity increment curve in data obtained by testing is selected as input at will, the corresponding 70% capacity value is used as output, and then the corresponding calculation formula is used for calculation.
In one embodiment, step 106 includes: inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into a trained random forest operator to obtain the output of each tree in the random forest; the output expression of the jth tree in the forest of the random forest operator is as follows:
Figure BDA0003670721480000111
wherein, c j Output of jth tree in forest representing random forest operator, X i It is indicated that the point (i) is,
Figure BDA0003670721480000112
represents a training sample set, Θ j It is meant that the random variables are distributed the same,
Figure BDA0003670721480000113
representing all of the cells containing x,
Figure BDA0003670721480000114
indicates all fall in
Figure BDA0003670721480000115
The number of points.
Adopting a bagging strategy according to the output of each tree in the random forest to obtain the target lithium ion battery capacity; the expression of the bagging strategy of the random forest operator is as follows:
Figure BDA0003670721480000116
wherein the content of the first and second substances,
Figure BDA0003670721480000117
the results of the random forest estimation are shown,
Figure BDA0003670721480000118
the estimation result of the jth tree in the random forest is shown, and M represents the number of trees.
It should be understood that, although the steps in the flowchart of fig. 1 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 1 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
In one specific example, the lithium ion battery capacity estimation results are shown in fig. 4 below, which is very similar to the measurements of all cells studied throughout the life cycle. The Root Mean Square Error (RMSE) of the estimated capacity was 3.91mAh (about 0.53% of the nominal capacity), and the Maximum Absolute Error (MAE) and the Average Absolute Error (AAE) were 11.93mAh and 3.12mAh, respectively. Compared with the traditional related method, the method has the advantages that the interference of voltage acquisition noise on the capacity increment curve can be effectively solved, and the difficulty of effectively extracting the characteristics of the capacity increment curve is solved. Meanwhile, in order to integrate the advantages of machine learning, a random forest algorithm is adopted to fuse the currently popular single learning machine: the battery capacity is estimated by a support vector machine, a long-time and short-time memory network and a Gaussian process regression algorithm, and the defect that a single machine learning algorithm is easy to fall into local optimization is effectively overcome.
In one embodiment, as shown in fig. 5, there is provided a lithium ion battery remaining capacity estimation apparatus including: the device comprises an experimental data acquisition module, a training sample determination module and a target lithium ion battery capacity determination module, wherein:
the test data acquisition module is used for acquiring a plurality of groups of measured values of a plurality of target lithium ion batteries in an aging test, wherein the measured values comprise the voltage, the current and the capacity of the target lithium ion batteries;
the training sample determining module is used for obtaining capacity increment data according to the voltage capacity relation obtained by fitting by adopting a fitting mode of a support vector machine according to the measured value and extracting a capacity increment peak of the capacity increment data; taking the plurality of capacity increment wave crests as training samples;
the target lithium ion battery capacity determining module is used for respectively inputting the training samples into the trained SVM, LSTM network and Gaussian process regression operator to respectively obtain a first predicted battery capacity, a second predicted battery capacity and a third predicted battery capacity; and inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into the trained random forest operator to obtain the target lithium ion battery capacity.
In one embodiment, the experimental data acquisition module is further configured to perform an experiment on a plurality of target lithium ion batteries in a predetermined test environment, and perform an aging experiment by repeatedly charging and discharging; in a cycle life test, after a preset aging period, a preset rated capacity value is subjected to charge-discharge cycle, the current capacity is calibrated, and data acquisition is carried out by adopting a preset sampling frequency, so that the voltage, the current and the capacity of the target lithium ion battery are obtained.
In one embodiment, the training sample determination module is further configured to obtain a voltage-to-capacity relationship by using the voltage of the measured value as an input and the capacity as an output and using a fitting mode of a support vector machine; calculating the voltage-capacity relation by adopting a capacity increment data calculation formula to obtain capacity increment data; the capacity increment data calculation formula is shown as formula (1). And extracting a capacity increment peak in the capacity increment data, and taking the capacity increment peak as a health characteristic parameter.
In one embodiment, the target lithium ion battery capacity determining module is further configured to input a training sample into a trained SVM, characterize the training sample into a high-dimensional spatial feature by using a kernel function, and obtain a first predicted battery capacity by using a nonlinear SVM for the high-dimensional spatial feature; inputting the training sample into a trained LSTM network, and capturing the long-term dependence relationship through a gate structure to obtain a second battery capacity; and inputting the training sample into a trained Gaussian process regression operator to obtain the third battery capacity.
In one embodiment, the target lithium ion battery capacity determining module is further configured to input the first predicted battery capacity, the second predicted battery capacity, and the third predicted battery capacity into a trained random forest operator to obtain an output of each tree in the random forest; the output expression of the jth tree in the forest of the random forest operator is shown as the formula (6); adopting a bagging strategy according to the output of each tree in the random forest to obtain the target lithium ion battery capacity; the expression of the bagging strategy of the random forest operator is shown in the formula (7).
For specific limitations of the lithium ion battery residual capacity estimation device, reference may be made to the above limitations of the lithium ion battery residual capacity estimation method, which will not be described herein again. All or part of the modules in the lithium ion battery residual capacity estimation device can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 6. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a lithium ion battery remaining capacity estimation method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 6 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In an embodiment, a computer device is provided, comprising a memory storing a computer program and a processor implementing the steps of the above method embodiments when executing the computer program.
In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (10)

1. A lithium ion battery residual capacity estimation method is characterized by comprising the following steps:
acquiring a plurality of groups of measured values of a plurality of target lithium ion batteries in an aging experiment; the measured values include voltage, current, and capacity of a target lithium ion battery;
adopting a fitting mode of a support vector machine according to the measured value, obtaining capacity increment data according to a voltage capacity relation obtained by fitting, and extracting a capacity increment peak of the capacity increment data;
taking a plurality of the capacity increment peaks as training samples;
respectively inputting the training samples into the trained SVM, LSTM network and Gaussian process regression operator to respectively obtain a first predicted battery capacity, a second predicted battery capacity and a third predicted battery capacity;
and inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into a trained random forest operator to obtain a target lithium ion battery capacity.
2. The method of claim 1, wherein obtaining multiple sets of measurements of the target lithium ion battery in an aging test comprises:
carrying out experiments on a plurality of target lithium ion batteries under a preset test environment, and carrying out aging experiments by repeated charge and discharge;
in the cycle life test, after a preset aging period, a preset rated capacity value is subjected to charge-discharge cycle, the current capacity is calibrated, and data acquisition is carried out by adopting a preset sampling frequency, so that the voltage, the current and the capacity of the target lithium ion battery are obtained.
3. The method of claim 1, wherein obtaining capacity increment data according to the voltage-capacity relationship obtained by fitting according to the measured values by using a fitting method of a support vector machine, and extracting a capacity increment peak of the capacity increment data comprises:
taking the voltage of the measured value as input and the capacity as output, and obtaining a voltage-capacity relation by adopting a fitting mode of a support vector machine;
calculating the voltage-capacity relation by adopting a capacity increment data calculation formula to obtain capacity increment data; the capacity increment data calculation formula is as follows:
Figure FDA0003670721470000011
wherein:
Figure FDA0003670721470000012
which represents the differential of the capacity versus the voltage,
Figure FDA0003670721470000013
representing the ratio of capacity increment to voltage increment, Q t 、Q t-1 Respectively representing the capacity values, V, at time t and t-1 t 、V t-1 Respectively representing the voltage values at the time t and t-1;
and extracting a capacity increment peak in the capacity increment data, and taking the capacity increment peak as a health characteristic parameter.
4. The method of claim 1, wherein inputting the training samples into trained SVM, LSTM network, and gaussian process regression operators to obtain a first predicted battery capacity, a second predicted battery capacity, and a third predicted battery capacity, respectively, comprises:
inputting the training samples into a trained SVM, characterizing the training samples into a high-dimensional space feature by adopting a kernel function, and obtaining a first predicted battery capacity by adopting a nonlinear SVM for the high-dimensional space feature;
inputting the training sample into a trained LSTM network, and capturing a long-term dependence relationship through a gate structure to obtain a second battery capacity;
and inputting the training sample into a trained Gaussian process regression operator to obtain a third battery capacity.
5. The method of claim 1, wherein inputting the first predicted battery capacity, the second predicted battery capacity, and the third predicted battery capacity into a trained random forest operator to obtain a target lithium ion battery capacity comprises:
inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into a trained random forest operator to obtain the output of each tree in the random forest; the output expression of the jth tree in the forest of the random forest operator is as follows:
Figure FDA0003670721470000021
wherein, c j Output of jth tree in forest representing random forest operator, X i It is indicated that the point (i) is,
Figure FDA0003670721470000022
represents a training sample set, Θ j It is meant that the random variables are distributed the same,
Figure FDA0003670721470000023
representing all of the cells containing x,
Figure FDA0003670721470000024
indicates all fall in
Figure FDA0003670721470000025
The number of points;
adopting a bagging strategy according to the output of each tree in the random forest to obtain the target lithium ion battery capacity; the expression of the bagging strategy of the random forest operator is as follows:
Figure FDA0003670721470000031
wherein the content of the first and second substances,
Figure FDA0003670721470000032
the results of the random forest estimation are shown,
Figure FDA0003670721470000033
the estimation result of the jth tree in the random forest is shown, and M represents the number of trees.
6. A lithium ion battery remaining capacity estimation apparatus, characterized in that the apparatus comprises:
the test data acquisition module is used for acquiring a plurality of groups of measured values of a plurality of target lithium ion batteries in an aging test, wherein the measured values comprise voltage, current and capacity of the target lithium ion batteries;
the training sample determining module is used for obtaining capacity increment data according to the voltage capacity relation obtained by fitting by adopting a fitting mode of a support vector machine according to the measured value and extracting a capacity increment peak of the capacity increment data; taking a plurality of the capacity increment peaks as training samples;
the target lithium ion battery capacity determining module is used for respectively inputting the training samples into the trained SVM, LSTM network and Gaussian process regression operator to respectively obtain a first predicted battery capacity, a second predicted battery capacity and a third predicted battery capacity; and inputting the first predicted battery capacity, the second predicted battery capacity and the third predicted battery capacity into a trained random forest operator to obtain a target lithium ion battery capacity.
7. The device of claim 1, wherein the experimental data acquisition module is further configured to perform an experiment on a plurality of target lithium ion batteries in a predetermined test environment, and perform an aging experiment by repeatedly charging and discharging; in a cycle life test, after a preset aging period, a preset rated capacity value is subjected to charge-discharge cycle, the current capacity is calibrated, and data acquisition is carried out by adopting a preset sampling frequency, so that the voltage, the current and the capacity of the target lithium ion battery are obtained.
8. The device of claim 1, wherein the training sample determining module is further configured to obtain a voltage-capacity relationship by using a fitting manner of a support vector machine with the voltage of the measured value as an input and the capacity as an output; calculating the voltage-capacity relation by adopting a capacity increment data calculation formula to obtain capacity increment data; the capacity increment data calculation formula is as follows:
Figure FDA0003670721470000034
wherein:
Figure FDA0003670721470000041
which represents the differential of the capacity versus the voltage,
Figure FDA0003670721470000042
representing the ratio of capacity increment to voltage increment, Q t 、Q t-1 Respectively representing the capacity values, V, at time t and t-1 t 、V t-1 Respectively representing the voltage values at the time t and t-1;
and extracting a capacity increment peak in the capacity increment data, and taking the capacity increment peak as a health characteristic parameter.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 6 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
CN202210604284.XA 2022-05-31 2022-05-31 Lithium ion battery residual capacity estimation method, device, equipment and storage medium Pending CN114814618A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210604284.XA CN114814618A (en) 2022-05-31 2022-05-31 Lithium ion battery residual capacity estimation method, device, equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210604284.XA CN114814618A (en) 2022-05-31 2022-05-31 Lithium ion battery residual capacity estimation method, device, equipment and storage medium

Publications (1)

Publication Number Publication Date
CN114814618A true CN114814618A (en) 2022-07-29

Family

ID=82519762

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210604284.XA Pending CN114814618A (en) 2022-05-31 2022-05-31 Lithium ion battery residual capacity estimation method, device, equipment and storage medium

Country Status (1)

Country Link
CN (1) CN114814618A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115389947A (en) * 2022-10-27 2022-11-25 北京航空航天大学 Lithium battery health state prediction method and device, electronic equipment and storage medium
CN116840721A (en) * 2023-06-02 2023-10-03 暨南大学 Lithium ion battery capacity estimation method, device, equipment and storage medium

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115389947A (en) * 2022-10-27 2022-11-25 北京航空航天大学 Lithium battery health state prediction method and device, electronic equipment and storage medium
CN115389947B (en) * 2022-10-27 2023-01-03 北京航空航天大学 Lithium battery health state prediction method and device, electronic equipment and storage medium
CN116840721A (en) * 2023-06-02 2023-10-03 暨南大学 Lithium ion battery capacity estimation method, device, equipment and storage medium

Similar Documents

Publication Publication Date Title
Li et al. Lithium-ion battery capacity estimation—A pruned convolutional neural network approach assisted with transfer learning
Li et al. Prognostic health condition for lithium battery using the partial incremental capacity and Gaussian process regression
Zhang et al. A novel method for lithium-ion battery state of energy and state of power estimation based on multi-time-scale filter
Tong et al. Early prediction of remaining useful life for Lithium-ion batteries based on a hybrid machine learning method
Cui et al. State of health diagnosis model for lithium ion batteries based on real-time impedance and open circuit voltage parameters identification method
Xiong et al. A data-driven multi-scale extended Kalman filtering based parameter and state estimation approach of lithium-ion polymer battery in electric vehicles
Liu et al. An integrated probabilistic approach to lithium-ion battery remaining useful life estimation
Dong et al. Lithium-ion battery state of health monitoring and remaining useful life prediction based on support vector regression-particle filter
Li et al. A mutated particle filter technique for system state estimation and battery life prediction
CN114814618A (en) Lithium ion battery residual capacity estimation method, device, equipment and storage medium
CN108872869B (en) Lithium ion battery degradation classification method based on BP neural network
Li et al. State‐of‐health prediction for lithium‐ion batteries via electrochemical impedance spectroscopy and artificial neural networks
Shi et al. A fast state-of-health estimation method using single linear feature for lithium-ion batteries
Luo et al. An online state of health estimation technique for lithium-ion battery using artificial neural network and linear interpolation
CN115201686B (en) Lithium ion battery health state assessment method under incomplete charge and discharge data
Zheng et al. State of health estimation for lithium battery random charging process based on CNN-GRU method
Luo et al. A simple feature extraction method for estimating the whole life cycle state of health of lithium-ion batteries using transformer-based neural network
Zhou et al. Research on online capacity estimation of power battery based on EKF-GPR model
Ang et al. Efficient linear predictive model with short term features for lithium-ion batteries state of health estimation
CN115935206A (en) SOC estimation method and device of energy storage system, equipment and storage medium
Hu et al. State of health estimation for lithium-ion batteries with dynamic time warping and deep kernel learning model
Li et al. Characterizing degradation in lithium-ion batteries with pulsing
CN113361692A (en) Lithium battery residual life combined prediction method
Densmore et al. Modeling the condition of lithium ion batteries using the extreme learning machine
Ghosh et al. An evolving quantum fuzzy neural network for online state-of-health estimation of Li-ion cell

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