CN114184972A - Method and equipment for automatically estimating SOH (state of health) of battery by combining data driving with electrochemical mechanism - Google Patents
Method and equipment for automatically estimating SOH (state of health) of battery by combining data driving with electrochemical mechanism Download PDFInfo
- Publication number
- CN114184972A CN114184972A CN202111285759.5A CN202111285759A CN114184972A CN 114184972 A CN114184972 A CN 114184972A CN 202111285759 A CN202111285759 A CN 202111285759A CN 114184972 A CN114184972 A CN 114184972A
- Authority
- CN
- China
- Prior art keywords
- soh
- curve
- battery
- voltage
- regression
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 72
- 230000007246 mechanism Effects 0.000 title claims abstract description 16
- 230000036541 health Effects 0.000 title description 13
- 239000000178 monomer Substances 0.000 claims abstract description 7
- 238000009499 grossing Methods 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 10
- 239000006185 dispersion Substances 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 8
- 238000010801 machine learning Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 description 9
- 230000032683 aging Effects 0.000 description 8
- 238000005259 measurement Methods 0.000 description 8
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 6
- 238000001914 filtration Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000003862 health status Effects 0.000 description 3
- 238000007477 logistic regression Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 238000007599 discharging Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000009830 intercalation Methods 0.000 description 2
- 230000002687 intercalation Effects 0.000 description 2
- 229910052744 lithium Inorganic materials 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000000691 measurement method Methods 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 241001123248 Arma Species 0.000 description 1
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000009831 deintercalation Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000003487 electrochemical reaction Methods 0.000 description 1
- 239000007772 electrode material Substances 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 229910002804 graphite Inorganic materials 0.000 description 1
- 239000010439 graphite Substances 0.000 description 1
- 238000011545 laboratory measurement Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 229910001416 lithium ion Inorganic materials 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
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]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
Abstract
The invention relates to a method and equipment for automatically estimating SOH of a battery by combining data driving and an electrochemical mechanism, which comprises the following steps: acquiring current, terminal voltage and temperature information of each battery cell, and calculating an IC curve, a DV curve, a voltage curve and a temperature difference curve in each charging period; respectively extracting key point sequences of all curves in a cycle period of the same charge-discharge mode; obtaining an SOH related index set based on the key point sequence of each curve and regression of the time sequence; obtaining the final SOH estimated SOH of each battery monomer based on the index weight acquired offline in advance and the SOH related index groupi(ii) a According to SOH of each battery celliAnd combining the SOH average level and distribution difference of the single body to obtain the final SOH estimation of the whole battery cluster. Compared with the prior art, the method has the advantages of effectiveness, accuracy and the like.
Description
Technical Field
The invention relates to the technical field of battery SOH estimation, in particular to a battery SOH automatic estimation method and device combining data driving and an electrochemical mechanism.
Background
The development Of an SOH (State Of Health, which refers to the capacity, Health and performance State Of a storage battery) online monitoring technology is more challenging due to a complex electrochemical mechanism in the battery aging process. Although the resistance growth problem can be evaluated by off-line testing and on-line identification algorithms, the detection of capacity fade still relies heavily on laboratory measurements and off-line analysis. SOH of battery cells has formed some effective methods, mainly classified into the following:
(1) starting from the definition of SOH, the direct measurement method estimates the SOH by measuring the battery capacity and the cycle number, and the method can obtain accurate SOH estimation but is difficult to apply online.
(2) One conventional and most common method of determining the battery capacity fade is based on an OCV-SOC curve. However, it requires that the battery be fully charged or discharged at a low rate, or that the open circuit voltage be measured after a long period of relaxation at a full range of SOC levels. Both methods require time-consuming testing and are therefore not suitable for the implementation of field operational data. There is a difficulty in directly performing SOH diagnosis on the battery cell based on the OCV. The intercalation and deintercalation of lithium ions only occur within a narrow voltage range. All voltage platforms lie within and overlap the critical voltage range, and the voltages are susceptible to measurement noise. All of these factors make the characterization of the OCV curve for most lithium battery chemistries difficult to directly observe. At present, many online SOH estimation methods comprehensively utilize or individually extract key feature points such as peak value, area and slope of IC, DV and DT curves to construct a health factor (set), and use SOH data obtained by an offline life test to establish an offline SOH estimation model. And then, extracting the health factor used in a single cycle on line, and calculating by using the established off-line SOH estimation model to obtain an on-line SOH estimation value. These methods are established on the assumption that the observed battery cell is consistent with the cell performance used by the offline model, however, although IC, DV, DT are proven to be effective tools for analyzing the battery capacity attenuation, since all the peaks on these curves are located in the voltage plateau region of the V-Q curve, which is relatively flat, they are sensitive to measurement noise, and small measurement errors cause the peak to change greatly. Furthermore, relying solely on an offline model for online SOH estimation is often unreliable due to differences in cell process consistency.
(3) Another method of estimating capacity loss is a calculation-based analysis method, including Incremental Capacity Analysis (ICA), Differential Voltage Analysis (DVA), and Differential Temperature (DT) analysis. By distinguishing between battery charge capacity (Q) and terminal voltage (V), the ICA converts the voltage plateau on the charge/discharge voltage (V-Q) curve, which segments the graphite anode to form a clearly identifiable dQ/dV peak on the Incremental Capacity (IC) curve. The concept of ICA is derived from the study of the lithium intercalation process and the corresponding phase phenomenon. The ICA has advantages in detecting gradual changes in battery behavior during life cycle tests, with higher sensitivity than that based on conventional charge and discharge curves, and generating key information on battery behavior related to electrochemical characteristics. The DVA method establishes a regression model by constructing the relationship between the external characteristics (mainly key characteristic points in a differential voltage curve) of the battery and the aging in an off-line manner so as to estimate the aging information of the battery on line. The technologies need to perform an off-line life test, establish aging libraries of different battery cores, different charging and discharging modes and working conditions, establish regression models of health factors and SOH, and find the corresponding regression models for online SOH estimation when in online use.
(4) The adaptive filtering algorithm represented by Kalman filtering, particle filtering and improved algorithms thereof is very suitable for solving the problem of battery nonlinear complex system state estimation, but an electrochemical model needs to be established to fit the electrochemical process, and the identification of various parameters in the model determines the model capability. However, filtering methods such as kalman filtering and particle filtering rely too much on the battery state model, and an appropriate battery degradation state space model is often not easily obtained in practice.
(6) The black box modeling is carried out by using relevant sample data through machine learning methods such as a support vector machine and a neural network, and the method has the advantages that a mathematical model and priori knowledge do not need to be established, and the mapping relation between the features and the SOH is directly learned from the data. However, the black box model based on machine learning needs to collect aging sample data and corresponding real tags, the collection of the aging sample data and the corresponding real tags consumes time, and the black box model is not feasible in many practical application scenes, and meanwhile, the battery consistency cannot be guaranteed, so that the black box model is difficult to be well applied in the practical field.
(7) Based on a Thevenin model battery life mathematical model, an SOH model is obtained through offline parameter identification, and SOH online estimation is achieved. The method has the advantages that the SOH is estimated by adopting the process data monitored by the BMS in real time, the requirement on equipment is low, unnecessary loss of the battery can not be generated, and the method is more suitable for the actual situation of online application. However, the Thevenin model-based algorithm makes several assumptions for simplifying model derivation, which does not conform to engineering practice and introduces certain errors, resulting in reduced estimation accuracy.
In addition, currently, the health state of the battery cells is generally evaluated, and it is difficult to directly obtain the health state of the battery cluster. In the prior art, the uniformity of the SOH of the single battery is also considered, and an SOH evaluation method of the battery pack is provided, but a fixed non-uniformity measurement method is adopted, which is biased to be accurate for different application scenes.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide an effective and accurate method and equipment for automatically estimating the SOH of a battery by combining data driving and an electrochemical mechanism.
The purpose of the invention can be realized by the following technical scheme:
a method for automatically estimating SOH of a battery by combining data driving with an electrochemical mechanism comprises the following steps:
acquiring current, terminal voltage and temperature information of each battery cell, and calculating an IC curve, a DV curve, a voltage curve and a temperature difference curve in each charging period;
respectively extracting key point sequences of all curves in a cycle period of the same charge-discharge mode;
based on the key point sequences of the curves and regression to the time series, a set of SOH-related indices (SOH) is obtainedIC_i,SOHDV_i,SOHV_i,SOHDT_i) Wherein, SOHIC_i、SOHDV_i、SOHV_i、SOHDT_iSOH obtained by predicting an IC curve, a DV curve, a voltage curve and a temperature difference curve on-line time sequence respectively;
index weight (omega) based on pre-offline acquisitionIC_i,ωDV_i,ωV_i,ωDT_i) And a stationObtaining the final SOH estimated SOH of each battery cell by the SOH related index setiWherein, ω isIC_i、ωDV_i、ωV_i、ωDT_iRespectively the weight of each curve;
according to SOH of each battery celliAnd combining the SOH average level and distribution difference of the single body to obtain the final SOH estimation of the whole battery cluster.
Further, for the IC curve, the key point extraction specifically is: and fitting the charging data by using Gaussian regression to obtain an IC curve, performing differential processing, and extracting a plurality of peak values of the IC curve as key points of the IC curve.
Further, for a DV curve, the key point extraction specifically is: and calculating the measured voltage and current data by adopting a numerical derivative method to obtain a DV curve, smoothing the DV curve by adopting second-order polynomial fitting, and extracting a plurality of peak values from the smoothed DV curve to be used as key points of the DV curve.
Further, for the voltage curve, the key point extraction specifically is as follows: identifying the inflection point of the voltage curve, acquiring a mathematically defined continuous function of the curvature of the inflection point, and taking a plurality of local minimum values of a standard closed form corresponding to the continuous function as key points of the voltage curve.
Further, for the temperature difference curve, the key point extraction specifically includes: and calculating the gradient of the temperature T of the battery core relative to time to obtain a temperature difference curve, preprocessing the DT-V curve by adopting a denoising smoothing method, and extracting peak values and valley values on the temperature difference curve as key points.
Further, a time series prediction method is adopted for regression of the time series, and the time series prediction method comprises a traditional machine learning regression model, a time series modeling method or a deep learning regression model.
Further, the index weight is obtained by regression through an off-line trained support vector regression SVR.
Further, the calculation formula of the final SOH estimate of the entire battery cluster is:
wherein the SOH is a final SOH estimate of the entire battery cluster,
mu is a discrete coefficient, and N is the number of electric cores in the cluster.
Further, the dispersion coefficient includes a dispersion relaxation coefficient, a range coefficient, a mean difference coefficient, or a standard deviation coefficient.
The present invention also provides an electronic device comprising:
one or more processors;
a memory; and
one or more programs stored in the memory, the one or more programs including instructions for performing the data-driven battery SOH auto-estimation method in combination with an electrochemical mechanism as described above.
Compared with the prior art, the invention has the following beneficial effects:
1) the invention provides an on-line dynamic SOH estimation technology which can better meet the actual engineering under the special requirement of the on-line SOH estimation problem of a battery cluster running in the actual working condition. An online SOH monitoring framework was developed by establishing quantitative relationships between battery capacity and IC peak, DV peak, surface temperature differential peak, voltage feature points. Compared with other online single SOH estimation methods, the method combines an offline model established by an aging test and online characteristic time sequence prediction to obtain the SOH estimation, and solves the problem that the battery process consistency is greatly influenced by singly depending on the offline model. Meanwhile, the comprehensive use of various health factors reduces the sensitivity of the estimation model to measurement errors, so that the reliability of practical application is improved.
2) The cluster SOH evaluation method provided by the invention measures the SOH difference of monomers in the cluster by using the dispersion with the relaxation coefficient, and reflects the average level of the SOH of the cluster.
3) The invention comprehensively considers various measurement curves and carries out denoising treatment, thereby being insensitive to measurement noise and having high SOH estimation precision.
4) The method can effectively extract key features related to the SOH of the battery, further improve the estimation precision and has high reliability.
5) The invention comprehensively utilizes various health factors to jointly estimate the SOH, improves the practicability of the algorithm and makes the algorithm more accord with the engineering practice.
6) The method utilizes the constructed health factor time series regression to obtain the actual characteristics of the local battery cell, integrates the offline model and the local characteristics, and reduces the estimation error of the offline model caused by the inconsistency of the battery cells.
7) The method is based on single SOH estimation, simultaneously measures the inconsistency of the SOH among the single SOH, measures the SOH of the cluster jointly from two angles of overall distribution and difference, can accurately and reasonably estimate the SOH of the cluster, provides more reliable performance estimation for the overall application based on the cluster, and has higher estimation accuracy and more comprehensive estimation.
Drawings
FIG. 1 is a schematic flow chart of the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
The inventors have found that the technique of SOH estimation for a single body cannot be directly applied to SOH estimation for clusters. As a result of intensive studies, it has been found that when monomers constitute a cluster, the overall health status of the cluster is influenced not only by the individual health status of all monomers constituting the cluster, but also by the difference in health status between the monomers. Therefore, the invention creatively provides the method for automatically estimating the SOH of the battery by combining data driving and an electrochemical mechanism, and the attenuation characteristics of single batteries and clusters can be accurately extracted on line.
Example 1
Referring to fig. 1, the present embodiment provides a method for automatically estimating SOH of a battery by combining data driving and an electrochemical mechanism, comprising the following steps:
acquiring current, terminal voltage and temperature information of each battery cell, and calculating an IC curve, a DV curve, a voltage curve and a temperature difference curve in each charging period;
preprocessing each curve, including denoising, smoothing and the like, and respectively extracting a key point sequence of each curve in a cycle period of the same charge-discharge mode;
based on the key point sequences of the curves and regression to the time series, a set of SOH-related indices (SOH) is obtainedIC_i,SOHDV_i,SOHV_i,SOHDT_i) Wherein, SOHIC_i、SOHDV_i、SOHV_i、SOHDT_iSOH obtained by predicting an IC curve, a DV curve, a voltage curve and a temperature difference curve on-line time sequence respectively;
index weight (omega) based on pre-offline acquisitionIC_i,ωDV_i,ωV_i,ωDT_i) And obtaining the final SOH estimated SOH of each battery monomer by the SOH related index setiWherein, ω isIC_i、ωDV_i、ωV_i、ωDT_iRespectively the weight of each curve;
according to SOH of each battery celliAnd combining the SOH average level and distribution difference of the single body to obtain the final SOH estimation of the whole battery cluster.
The key points of each curve are extracted by the following method:
1) IC curve
The IC curve is represented as dQ/dV, where Q is the instantaneous charge capacity and V is the terminal voltage of the battery. There are multiple peaks in the IC curve, with each peak representing a particular voltage plateau. The peak characteristics are directly related to the phase change of the electrode material and show a high sensitivity to the aging of the battery. In this embodiment, a gaussian regression is used to fit the charging data to obtain an IC curve, and a differentiation process is performed to extract a plurality of peak values of the IC curve as key points of the IC curve. Three peak values are extracted as feature key points in the embodiment.
2) Voltage curve
In a complete charging and discharging curve, the voltage shows a stepwise change due to different stages of the electrochemical reaction process of the battery cell. The invention adopts an inflection point identification algorithm to extract voltage key characteristic points in a single charge-discharge cycle. A mathematical definition of curvature is used as a continuous function of the basis for the voltage critical characteristic point definition. For any continuous function f, there is a standard closed form Kf(x) It defines the curvature of f at any point as a function first and second derivatives. The key point is defined as Kf(x) And extracting a plurality of local minimum values as key points of the voltage curve. In this embodiment, three local minimum values are extracted as feature key points.
Standard closed form Kf(x) The expression of (a) is:
3) DV Curve
The DV curve is calculated as the gradient of V versus Q: dV/dQ, and obtaining a differential curve by adopting a numerical derivative method according to measured voltage and current data. The measurement data may be expressed as:
in order to remove the influence of measurement noise, second-order polynomial fitting is adopted, and then a window is moved to smooth the DV curve to obtain a new DV curve. And extracting a plurality of peak values on the DV curve to form characteristic key points of the DV curve. Two peak values are extracted as feature key points in the embodiment.
4) Temperature difference curve (DT curve)
The DT curve calculates the gradient of the temperature T with respect to time:
forming DT-V curve, preprocessing the DT-V curve by denoising smoothing method, and extracting DT curveThe peak and valley values of (a) are taken as characteristic key points.
The denoising and smoothing mode of each curve can adopt any denoising and smoothing method in the prior art.
Extracting cycle periods of the same charge-discharge mode from the battery cell operation monitoring data, marking corresponding charge-discharge characteristics and cycle period serial numbers, and reconstructing the extracted voltage characteristic points according to the cycle period marks and the sequence of the characteristic points in a single charge-discharge cycle to obtain a voltage key characteristic point sequence { V }11，V12,…V1n},{V21，V22,…V2n},{V31，V32,…V3nIC curve key point sequence IC1：n＝{IC11，IC12,…IC1n},{IC21，IC22,…IC2n},{IC31，IC32,…IC3nDV curve key point sequence DV1：n＝{DV11，DV12,…DV1n},{DV21，DV22,…DV2n},{DV31，DV32,…DV3nKey point sequence DT of DT curve1：n＝{DT11，DT12,…DT1n},{DT21，DT22,…DT2n}。
After the key point sequences of the curves are obtained, estimating the SOH estimation value of each battery cell in the current cycle period by an online time sequence prediction method, specifically:
for IC curve key point sequence IC1：n＝{IC11，IC12,…IC1n},{IC21，IC22,…IC2n},{IC31，IC32,…IC3nDenoising and smoothing to obtain an IC curve key point sequence IC'1：n＝{IC’11，IC’12,…IC’1n},{IC’21，IC’22,…IC’2n},{IC’31，IC’32,…IC’3n}. Using traditional machine learning regression model (such as linear regression), sequential modeling method ARMA/ARIMA, or deep learning regression model RNN, LSTM, etcMethod for predicting time series by using equal time series and establishing time series prediction model fICIf the SOH estimated value of the cell i in the current cycle period is SOHIC_i＝fIC(IC’n)；
For DV curve key point sequence DV1：n＝{DV11，DV12,…DV1n},{DV21，DV22,…DV2n},{DV31，DV32,…DV3nDenoising and smoothing are carried out to obtain a DV curve key point sequence DV'1：n＝{DV’11，DV’12,…DV’1n},{DV’21，DV’22,…DV’2n},{DV’31，DV’32,…DV’3n}. Establishing a time sequence prediction model f by adopting time sequence prediction methods such as logistic regression, AR, LSTM and the likeDVThen cell i SOH at the current cycleDV_i＝fDV(DV’n)；
For voltage curve key point sequence V1：n＝{V11，V12,…V1n},{V21，V22,…V2n},{V31，V32,…V3nDenoising and smoothing are carried out to obtain a V curve key point sequence V'1：n＝{V’11，V’12,…V’1n},{V’21，’22,…V’2n},{V’31，V’32,…V’3n}. Establishing a time sequence prediction model f by adopting time sequence prediction methods such as logistic regression, AR, LSTM and the likeVThen cell i SOH at the current cycleV_i＝fV(V’n)；
For DT curve key point sequence DV1：n＝{DT11，DT12,…DT1n},{DT21，DT22,…DT2nDenoising and smoothing are carried out to obtain a DT curve key point sequence DT'1：n＝{DT’11，DT’12,…DT’1n},{DT’21，DT’22,…DT’2n}. Establishing time sequence by adopting time sequence prediction methods such as logistic regression, AR, LSTM and the likePrediction model fDTThen cell i SOH at the current cycleDT_i＝fDT(DT’n)；
Obtaining a set of SOH-related indicators (SOH) based on the above processIC_i,SOHDV_i,SOHV_i,SOHDT_i) Combined with pre-offline acquired index weights (ω)IC_i,ωDV_i,ωV_i,ωDT_i) Then the SOH of the cell i can be obtained by calculationiAnd (6) estimating. In the present embodiment, the index weight is represented as ω1、ω2、ω3、ω4Then SOH of cell iiThe estimate is expressed as:
wherein the content of the first and second substances,
ωithe support vector regression SVR after off-line training is adopted for regression, and the support vector regression SVR can be replaced by other machine learning models (such as various ensemble learning models, decision trees and neural networks) and SVR improved models.
In this embodiment, the SVR regression model is as follows:
assuming that a sample set { xi, yi }, xi is equal to Rn, i is equal to 1,2, …, and n, xi are feature sets; yi is the corresponding sample, then the equation for the optimal classification plane is:
y＝ωTΦ(x)+b (4)
wherein ω is a weight vector; b is an offset vector.
For this purpose, the relaxation factor is used to transform the data, and the following equivalent problem is obtained:
wherein C is a penalty parameter.
In order to simplify the problem solving process and accelerate the solving speed, the Lagrange multiplier α i is introduced to obtain the following even problem:
the formula for ω can be obtained from the above equation:
ω＝∑αiyiΦ(xi)·Φ(x) (7)
the classification method for obtaining the SVM is as follows:
f(x)＝sgn(αiyiΦ(xi)·Φ(x)+b) (8)
Φ(xi) Phi (x) is a dot product operation, and is easy to cause the problem of dimension disaster, and a kernel function k (x, x) is adopted for the problemi) Instead of phi (x)i) Φ (x), thus becoming:
f(x)＝sgn(αiyik(x,xi)+b) (9)
the radial basis kernel function is chosen such that the above equation becomes:
assuming that the cluster is formed by connecting N electric cores in series, the SOH of the electric core i is the health stateiThe SOH of the cluster is calculated according to equation (11):
wherein the SOH is a final SOH estimate of the entire battery cluster,
mu is a discrete coefficient, and N is the number of electric cores in the cluster. The dispersion coefficient may be a dispersion relaxation coefficient, a range coefficient, a mean difference coefficient, or a standard deviation coefficient. In this example, μ is a dispersion relaxation coefficient, and μ ∈ [0,1 ]]。
In other embodiments, the keypoint time series regression process may be replaced by other machine learning regression models (e.g., classes of ensemble learning models, neural networks, SVRs, etc.).
In other embodiments, the method for combining the key feature points of different charging depth intervals may be replaced by a set of key points of multiple charging intervals, so long as at least one key point in the above invention is included in the set, and at most all key points are included in the set, depending on the charging depth and the start-stop charging state.
In other embodiments, the variance term in the cluster SOH calculation formula may be replaced by other monotonic functions with a value range of [1, n ].
The above method, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Example 2
The present embodiments provide an electronic device comprising one or more processors, memory, and one or more programs stored in the memory, the one or more programs including instructions for performing the method for automatic estimation of battery SOH in combination with data-driven and electrochemical mechanisms as described in embodiment 1.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (10)
1. A method for automatically estimating SOH of a battery by combining data driving and an electrochemical mechanism is characterized by comprising the following steps:
acquiring current, terminal voltage and temperature information of each battery cell, and calculating an IC curve, a DV curve, a voltage curve and a temperature difference curve in each charging period;
respectively extracting key point sequences of all curves in a cycle period of the same charge-discharge mode;
based on the key point sequences of the curves and regression to the time series, a set of SOH-related indices (SOH) is obtainedIC_i，SOHDV_i，SOHV_i，SOHDT_i) Wherein, SOHIC_i、SOHDV_i、SOHV_i、SOHDT_iSOH obtained by predicting an IC curve, a DV curve, a voltage curve and a temperature difference curve on-line time sequence respectively;
index weight (omega) based on pre-offline acquisitionIC_i，ωDV_i，ωV_i，ωDT_i) And obtaining the final SOH estimated SOH of each battery monomer by the SOH related index setiWherein, ω isIC_i、ωDV_i、ωV_i、ωDT_iRespectively the weight of each curve;
according to SOH of each battery celliAnd combining the SOH average level and distribution difference of the single body to obtain the final SOH estimation of the whole battery cluster.
2. The method for automatically estimating SOH of a battery by combining data driving and an electrochemical mechanism according to claim 1, wherein for an IC curve, the key point extraction is specifically as follows: and fitting the charging data by using Gaussian regression to obtain an IC curve, performing differential processing, and extracting a plurality of peak values of the IC curve as key points of the IC curve.
3. The data-driven electrochemical mechanism-combined battery SOH automatic estimation method according to claim 1, characterized in that, for DV curves, the key point extraction is specifically: and calculating the measured voltage and current data by adopting a numerical derivative method to obtain a DV curve, smoothing the DV curve by adopting second-order polynomial fitting, and extracting a plurality of peak values from the smoothed DV curve to be used as key points of the DV curve.
4. The method for automatically estimating SOH of a battery by combining data driving and an electrochemical mechanism according to claim 1, wherein for a voltage curve, the key point extraction is specifically as follows: identifying the inflection point of the voltage curve, acquiring a mathematically defined continuous function of the curvature of the inflection point, and taking a plurality of local minimum values of a standard closed form corresponding to the continuous function as key points of the voltage curve.
5. The method for automatically estimating SOH of a battery by combining data driving and an electrochemical mechanism according to claim 1, wherein for a temperature difference curve, the key point extraction is specifically as follows: and calculating the gradient of the temperature T of the battery core relative to time to obtain a temperature difference curve, preprocessing the DT-V curve by adopting a denoising smoothing method, and extracting peak values and valley values on the temperature difference curve as key points.
6. The method of claim 1 in which the time series prediction method is used to perform time series regression, and the time series prediction method comprises a conventional machine learning regression model, a time series modeling method or a deep learning regression model.
7. The method of claim 1, wherein the indicator weights are derived by regression using an off-line trained Support Vector Regression (SVR).
8. The method of claim 1 in which the equation for the final SOH estimate of the entire cluster is:
wherein the SOH is a final SOH estimate of the entire battery cluster,
mu is a discrete coefficient, and N is the number of electric cores in the cluster.
9. The method of claim 8 where the dispersion coefficient comprises a dispersion relaxation coefficient, a range coefficient, a mean difference coefficient, or a standard deviation coefficient.
10. An electronic device, comprising:
one or more processors;
a memory; and
one or more programs stored in the memory, the one or more programs including instructions for performing the method for data-driven automatic estimation of battery SOH in combination with an electrochemical mechanism according to any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111285759.5A CN114184972B (en) | 2021-11-02 | 2021-11-02 | Automatic estimation method and equipment for SOH of battery by combining data driving and electrochemical mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111285759.5A CN114184972B (en) | 2021-11-02 | 2021-11-02 | Automatic estimation method and equipment for SOH of battery by combining data driving and electrochemical mechanism |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114184972A true CN114184972A (en) | 2022-03-15 |
| CN114184972B CN114184972B (en) | 2023-12-22 |
Family
ID=80601776
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111285759.5A Active CN114184972B (en) | 2021-11-02 | 2021-11-02 | Automatic estimation method and equipment for SOH of battery by combining data driving and electrochemical mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114184972B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024023211A1 (en) * | 2022-07-28 | 2024-02-01 | Totalenergies Onetech | A method for characterizing the evolution of state of health of a device with duration of operation |
| CN117574788A (en) * | 2024-01-17 | 2024-02-20 | 中国第一汽车股份有限公司 | Multi-scale modeling-based battery health degree prediction method and device |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150066406A1 (en) * | 2013-08-27 | 2015-03-05 | The Regents Of The University Of Michigan | On-board state of health monitoring of batteries using incremental capacity analysis |
| CN105158698A (en) * | 2015-08-28 | 2015-12-16 | 江苏大学 | Battery pack state of health online estimation method based on charging voltage curve |
| CN106997028A (en) * | 2017-05-19 | 2017-08-01 | 江苏大学 | A kind of battery differential voltage and incremental capacity plot Efficient Solution method |
| CN111398833A (en) * | 2020-03-13 | 2020-07-10 | 浙江大学 | Battery health state evaluation method and evaluation system |
| CN111398837A (en) * | 2020-04-01 | 2020-07-10 | 重庆大学 | Vehicle battery health state estimation method based on data driving |
| CN111443293A (en) * | 2020-03-30 | 2020-07-24 | 青岛大学 | Lithium battery state of health (SOH) estimation method based on data driving |
| CN111948546A (en) * | 2020-08-07 | 2020-11-17 | 国网陕西省电力公司汉中供电公司 | Lithium battery health degree evaluation method and system |
| CN111965560A (en) * | 2020-08-24 | 2020-11-20 | 重庆大学 | Battery health state estimation method oriented to general discharge working condition |
| WO2020251854A1 (en) * | 2019-06-14 | 2020-12-17 | Cummins Inc. | Methods and devices for determining battery state of health using incremental capacity analysis and support vector regression |
| CN112158105A (en) * | 2020-10-12 | 2021-01-01 | 东风汽车集团有限公司 | SOH (State of health) evaluation device, method and system for power battery of new energy automobile |
| CN112485693A (en) * | 2020-11-19 | 2021-03-12 | 上海电力大学 | Battery health state rapid evaluation method based on temperature probability density function |
| US20210116513A1 (en) * | 2019-05-17 | 2021-04-22 | Contemporary Amperex Technology Co., Limited | Method for correcting soh, apparatus, battery management system and storage medium |
| CN112798960A (en) * | 2021-01-14 | 2021-05-14 | 重庆大学 | Battery pack residual life prediction method based on migration deep learning |
| CN113466696A (en) * | 2021-05-17 | 2021-10-01 | 北京交通大学 | Battery pack monomer state estimation method based on voltage curve transformation |
| US20210311129A1 (en) * | 2020-04-03 | 2021-10-07 | Cummins Inc. | Methods and systems for accelerated determining of state of health using incremental capacity analysis |
-
2021
- 2021-11-02 CN CN202111285759.5A patent/CN114184972B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150066406A1 (en) * | 2013-08-27 | 2015-03-05 | The Regents Of The University Of Michigan | On-board state of health monitoring of batteries using incremental capacity analysis |
| CN105158698A (en) * | 2015-08-28 | 2015-12-16 | 江苏大学 | Battery pack state of health online estimation method based on charging voltage curve |
| CN106997028A (en) * | 2017-05-19 | 2017-08-01 | 江苏大学 | A kind of battery differential voltage and incremental capacity plot Efficient Solution method |
| US20210116513A1 (en) * | 2019-05-17 | 2021-04-22 | Contemporary Amperex Technology Co., Limited | Method for correcting soh, apparatus, battery management system and storage medium |
| WO2020251854A1 (en) * | 2019-06-14 | 2020-12-17 | Cummins Inc. | Methods and devices for determining battery state of health using incremental capacity analysis and support vector regression |
| CN111398833A (en) * | 2020-03-13 | 2020-07-10 | 浙江大学 | Battery health state evaluation method and evaluation system |
| CN111443293A (en) * | 2020-03-30 | 2020-07-24 | 青岛大学 | Lithium battery state of health (SOH) estimation method based on data driving |
| CN111398837A (en) * | 2020-04-01 | 2020-07-10 | 重庆大学 | Vehicle battery health state estimation method based on data driving |
| US20210311129A1 (en) * | 2020-04-03 | 2021-10-07 | Cummins Inc. | Methods and systems for accelerated determining of state of health using incremental capacity analysis |
| CN111948546A (en) * | 2020-08-07 | 2020-11-17 | 国网陕西省电力公司汉中供电公司 | Lithium battery health degree evaluation method and system |
| CN111965560A (en) * | 2020-08-24 | 2020-11-20 | 重庆大学 | Battery health state estimation method oriented to general discharge working condition |
| CN112158105A (en) * | 2020-10-12 | 2021-01-01 | 东风汽车集团有限公司 | SOH (State of health) evaluation device, method and system for power battery of new energy automobile |
| CN112485693A (en) * | 2020-11-19 | 2021-03-12 | 上海电力大学 | Battery health state rapid evaluation method based on temperature probability density function |
| CN112798960A (en) * | 2021-01-14 | 2021-05-14 | 重庆大学 | Battery pack residual life prediction method based on migration deep learning |
| CN113466696A (en) * | 2021-05-17 | 2021-10-01 | 北京交通大学 | Battery pack monomer state estimation method based on voltage curve transformation |
Non-Patent Citations (2)
| Title |
|---|
| 姜久春;马泽宇;李雪;张彩萍;张维戈;: "基于开路电压特性的动力电池健康状态诊断与估计", 北京交通大学学报 * |
| 王振新;秦鹏;康健强;王菁;朱国荣;向馗;: "基于衰退机理的三元锂离子电池SOH的诊断与估算", 电子测量技术 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024023211A1 (en) * | 2022-07-28 | 2024-02-01 | Totalenergies Onetech | A method for characterizing the evolution of state of health of a device with duration of operation |
| CN117574788A (en) * | 2024-01-17 | 2024-02-20 | 中国第一汽车股份有限公司 | Multi-scale modeling-based battery health degree prediction method and device |
| CN117574788B (en) * | 2024-01-17 | 2024-05-14 | 中国第一汽车股份有限公司 | Multi-scale modeling-based battery health degree prediction method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114184972B (en) | 2023-12-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111707951B (en) | Battery pack consistency evaluation method and system | |
| Wei et al. | Remaining useful life prediction and state of health diagnosis for lithium-ion batteries using particle filter and support vector regression | |
| Zhang et al. | A rapid online calculation method for state of health of lithium-ion battery based on coulomb counting method and differential voltage analysis | |
| Wang et al. | State of health estimation based on modified Gaussian process regression for lithium-ion batteries | |
| EP3721499B1 (en) | Method and system for online assessing state of health of a battery | |
| CN110133525B (en) | Lithium ion battery health state estimation method applied to battery management system | |
| Xiong et al. | A data-driven method for extracting aging features to accurately predict the battery health | |
| CN114184972B (en) | Automatic estimation method and equipment for SOH of battery by combining data driving and electrochemical mechanism | |
| Fan et al. | A remaining capacity estimation approach of lithium-ion batteries based on partial charging curve and health feature fusion | |
| CN113447828B (en) | Lithium battery temperature estimation method and system based on Bayesian neural network | |
| Cai et al. | An estimation model for state of health of lithium-ion batteries using energy-based features | |
| Wei et al. | Lyapunov-based state of charge diagnosis and health prognosis for lithium-ion batteries | |
| CN114167284B (en) | Lithium battery RUL prediction method and equipment based on BMS big data and integrated learning | |
| CN113030761A (en) | Method and system for evaluating health state of battery of super-large-scale energy storage power station | |
| CN114839536A (en) | Lithium ion battery health state estimation method based on multiple health factors | |
| CN114035095B (en) | Lithium battery SOH estimation method, medium and equipment based on voltage curve inflection point identification | |
| CN112485695A (en) | Detection method and device for power battery | |
| CN115754724A (en) | Power battery state of health estimation method suitable for future uncertainty dynamic working condition discharge | |
| Lin et al. | Estimating the state of health of lithium-ion batteries based on a probability density function | |
| CN113820615A (en) | Battery health degree detection method and device | |
| Zhao et al. | Battery degradation stage detection and life prediction without accessing historical operating data | |
| CN117805649A (en) | Method for identifying abnormal battery cells based on SOH quantized battery capacity attenuation | |
| Guo et al. | Rapid online health estimation for lithium-ion batteries based on partial constant-voltage charging segment | |
| Santoni et al. | A guide to equivalent circuit fitting for impedance analysis and battery state estimation | |
| CN113447821B (en) | Method for evaluating state of charge of battery |
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 |