CN109871881A - A kind of battery performance analysis method based on agglomerate layered clustering algorithm - Google Patents
A kind of battery performance analysis method based on agglomerate layered clustering algorithm Download PDFInfo
- Publication number
- CN109871881A CN109871881A CN201910066652.8A CN201910066652A CN109871881A CN 109871881 A CN109871881 A CN 109871881A CN 201910066652 A CN201910066652 A CN 201910066652A CN 109871881 A CN109871881 A CN 109871881A
- Authority
- CN
- China
- Prior art keywords
- battery
- class
- operating condition
- curve
- clustered
- 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
Landscapes
- Secondary Cells (AREA)
Abstract
The present invention proposes a kind of battery performance analysis method based on agglomerate layered clustering algorithm, belongs to battery technology field.This method selects battery operating condition clustering factor X first, obtains to be analyzed operating condition of the measurement data as the battery of the X of battery to be analyzed；Then fundamental curve unit is converted element to be clustered by the fundamental curve unit for extracting clustering factor X in operating condition to be analyzed；It using agglomerate layered clustering algorithm, treats cluster element and is clustered, obtain battery at every kind and cluster the operation curve corresponded under operating condition.Utilize the method for the present invention, it may not need selection characteristic parameter, clustering directly is carried out to the performance curve in battery actual moving process, cluster analysis result can be the rate of decay of estimation battery actual motion, battery system configuration and control strategy are improved, ensures that cell safety, long-lived operation provide guidance.
Description
Technical field
The present invention relates to battery technology fields, and in particular to a kind of battery performance analysis based on agglomerate layered clustering algorithm
Method.
Background technique
Lithium ion battery has many advantages, such as that energy density is high, has extended cycle life, self-discharge rate is low, memory-less effect, becomes
The main driving power of pure electric automobile, plug-in electromobile and hybrid vehicle, is also applied to energy-storage system etc.
Several scenes.Battery actual operating mode under various application scenarios is analyzed, can be construction standard work under laboratory condition
Condition, approximate simulation actual operating mode carries out Cell Experimentation An, evaluation battery performance provides basis；Meanwhile battery actual motion work
The analysis of condition can also be labs cell degradation Acceleration study, and then estimate the rate of decay in battery actual moving process
It provides and supports；To the configuration and control strategy for improving battery system, to ensure that it is highly important that cell safety, long-lived operation have
Technical meaning.
The current existing analysis means for battery operating condition, main method are to extract battery operation operating mode feature parameter,
Including maximum value, minimum value, average value, standard deviation, duration etc., then by principal component analysis and clustering algorithm to characteristic parameter into
Row analysis.But the validity of this method only passes through the feature of battery operation operating condition very dependent on the selection of characteristic parameter
Parameter directly can not intuitively reflect the shape of battery operation performance curve；Thus need it is a kind of without characteristic parameter,
The method that clustering directly is carried out to battery operation performance curve.
Summary of the invention
The purpose of the present invention is places in order to overcome the deficiencies of the prior art, propose a kind of based on agglomerate layered clustering algorithm
Battery performance analysis method.Using the method for the present invention, selection characteristic parameter may not need, directly in battery actual moving process
Performance curve carry out clustering, cluster analysis result can for estimation battery actual motion rate of decay, improve battery system
System configuration and control strategy ensure that cell safety, long-lived operation provide guidance.
The present invention proposes a kind of battery performance analysis method based on agglomerate layered clustering algorithm, which is characterized in that the party
Method the following steps are included:
1) select battery operating condition clustering factor X, obtain the measurement data of the X of battery to be analyzed as the battery to
Analyze operating condition；The measurement data constitutes curve of the X of battery to be analyzed about time t；
2) the fundamental curve unit of clustering factor X in operating condition to be analyzed is extracted；
3) element to be clustered is converted by the fundamental curve unit that step 2) is extracted；
4) agglomerate layered clustering algorithm is utilized, cluster element is treated and is clustered, obtains battery in the corresponding work of every kind of cluster
Operation curve under condition.
The features of the present invention and beneficial effect are:
Using method of the invention, clustering directly can be carried out to the performance curve in battery actual moving process,
And then can be to construct standard condition under laboratory condition, approximate simulation actual operating mode, development Cell Experimentation An, evaluation are cell performance
Basis can be provided；It can also be labs cell degradation Acceleration study, and then estimate the decaying in battery actual moving process
It provides and supports；For improving the configuration and control strategy of battery system, to ensure that it is particularly significant that cell safety, long-lived operation have
Technical meaning.The present invention can be applicable to the fields such as electric car, energy storage.
Detailed description of the invention
Fig. 1 is the overall flow figure of the method for the present invention.
Fig. 2 is the measurement data schematic diagram that the vehicle battery pack current period is tested in the embodiment of the present invention.
Fig. 3 is that interpolation supplements zero point and extracts fundamental curve cell schematics in the embodiment of the present invention.
Fig. 4 is to convert element schematic diagram to be clustered for fundamental curve unit in the embodiment of the present invention.
Fig. 5 is element defined in the embodiment of the present invention apart from schematic diagram.
Fig. 6 is battery operating condition agglomerate layered cluster result schematic diagram in the embodiment of the present invention.
Specific embodiment
The present invention proposes a kind of battery performance analysis method based on agglomerate layered clustering algorithm, below in conjunction with attached drawing and tool
The present invention is described in detail for body embodiment.
The present invention proposes a kind of battery performance analysis method based on agglomerate layered clustering algorithm, overall flow such as Fig. 1 institute
Show, comprising the following steps:
1) select battery operating condition clustering factor X, obtain the measurement data of the X of battery to be analyzed as the battery to
Analyze operating condition；The measurement data constitutes curve of the X of battery to be analyzed about time t.
2) the fundamental curve unit of clustering factor X in operating condition to be analyzed is extracted；
3) element to be clustered is converted by the fundamental curve unit that step 2) is extracted；
4) agglomerate layered clustering algorithm is utilized, cluster element is treated and is clustered, obtains battery in the corresponding work of every kind of cluster
Operation curve under condition.
Battery operating condition clustering factor X, X is selected in the step 1) to be chosen as electric current I, voltage V, appointing in power P
It anticipates one kind, can voluntarily be selected according to actual conditions such as analysis purposes.The battery operating condition clustering factor X selected in the present embodiment
For battery pack current I.It is as shown in Figure 2 that the measurement data of vehicle battery pack current I period is tested in the embodiment of the present invention.Figure
2 illustrate the result conduct signal of 800s measurement data before the battery pack；It is a length of when the total data of actual analysis in the present embodiment
10h。
The present invention claims measurement data to include at least all fortune of one, battery to be analyzed workaday clustering factor X
Row data.
The step 2) extracts the fundamental curve unit of clustering factor X in operating condition to be analyzed, and the specific method is as follows: mentioning
Take X-t (time) curved section in operating condition to be analyzed between every two adjacent X zero point as the substantially bent of clustering factor X
Line unit.If due to the discreteness of floor data acquisition, in measurement data between adjacent, X value opposite sign two data points not
The data point (zero point) for being directly zero comprising X value, then can be by handling two adjacent data point interpolations, between supplement
Zero point.
The schematic diagram that the fundamental curve unit of clustering factor X is extracted in the present embodiment is as shown in Figure 3.In the present embodiment
Clustering factor X be battery pack current I；Due to floor data acquisition discreteness, it is adjacent, in Fig. 3 I value opposite sign two
It does not directly include zero point between a consecutive number strong point, thus to interpolation supplement zero between adjacent, I value opposite sign two data points
Point, shown in the data point marked such as the circle in Fig. 3.Supplement zero point after, fundamental curve unit be two I zero points closed on it
Between I-t curved section, as shown in the curved section in Fig. 3 in dotted line frame.
Element to be clustered is converted by the fundamental curve unit that step 2) is extracted in the step 3), the specific method is as follows:
Each fundamental curve unit that step 2) is extracted is converted into corresponding XdY profile is as element to be clustered.Method for transformation
Are as follows:
For each fundamental curve unit,
Wherein, if X (t) > 0 in fundamental curve unit,
If X (t) < 0 in fundamental curve unit,
Wherein, X (t) represents the corresponding X value of t moment, t in fundamental curve unit X-t curved sectionsFor fundamental curve unit X-
The start time point of t curved section, teFor the termination time point of fundamental curve unit X-t curved section, g (Xd, t) and it is intermediate computations letter
Number, XdFor operation variable.
By the calculation method of curve conversion it is found that Y should be the integration amount of X；In the present embodiment, then Y represents electricity Q；It is bent
Line conversion converts element I to be clustered for fundamental curve unit I-t curved- Q curve.In the present embodiment, due to I-t curve
It comes in every shape, if directly carrying out clustering for I-t curve as element to be clustered, clustering effect may be influenced；Thus
First convert I-t curve to dull Id- Q curve is as element to be clustered.Schematic diagram such as Fig. 4 that curve converts in the present embodiment
It is shown.The formula of calculation method is converted it is found that taking a current value I according to curved1(Id1>=0), corresponding Q1As I-t curve with
Straight line I=Id1The sum of each section area surrounded ∑ S, as shown in Figure 4；And Id- Q curve should be the curve of monotone decreasing, Id-Q
Maximum current I in curvemaxIt is equal with the maximum current in I-t curve；IdMaximum electricity Q in-Q curvemaxAs I-t is bent
The gross area that line and x-axis surround.In the present embodiment, 285 elements to be clustered are converted by 285 fundamental curve units.
Specific step is as follows for the step 4):
4-1) it regard each element to be clustered as one kind independent；
4-2) calculate the class distance between any two class；The specific method is as follows:
For any two class, if class A hasTotal NAA element, class B haveTotal NBA element, then class A
Class distance DA, B between class B are defined as " average distance " of two class all elements, it may be assumed that
Wherein, d (li,l'j) it is element liWith element ljBetween element distance；
Element distance definition is two element (X to be clustereddY profile) figure that surrounds with axis of abscissas and axis of ordinates
The gross area.In the present embodiment, element distance definition is two element (I to be clusteredd- Q curve) with axis of abscissas and vertical seat
The gross area for the figure that parameter surrounds, as shown in figure 5, two kinds of situation explanations can be specifically divided into.If two Id- Q curve liAnd ljPhase
It hands over, element distance d (l between the twoi, lj) it is the sum of area of two curved line trangles shown in shade ∑ S in figure；If two
Id- Q curve liAnd ljNon-intersecting, between the two element distance d (li,lj) be curvilinear boundary quadrilateral shown in shade in figure face
Product S.
4-3) using step 4-2's) as a result, this two class is merged into a new class apart from the smallest two class by selection class；
4-4) repeat step 4-2) to 4-3), until the sum of cluster reaches setting value N, the value range of N 4-25 it
Between, cluster finishes, and obtains battery at every kind and clusters the operation curve corresponded under operating condition, analysis finishes.
In the present embodiment, the agglomerate layered clustering method in step 4), the setting value of cluster sum N is 6, final to cluster
As a result the operation curve corresponded under operating condition is clustered at every kind as shown in fig. 6, battery can be obtained.It is every one kind in comprising it is several to
Cluster element (Id- Q curve), and the element distance in same class is closer to, and shows that its corresponding operating condition is similar.
Claims (5)
1. a kind of battery performance analysis method based on agglomerate layered clustering algorithm, which is characterized in that this method includes following step
It is rapid:
1) battery operating condition clustering factor X is selected, obtains the measurement data of the X of battery to be analyzed as the to be analyzed of the battery
Operating condition；The measurement data constitutes curve of the X of battery to be analyzed about time t；
2) the fundamental curve unit of clustering factor X in operating condition to be analyzed is extracted；
3) element to be clustered is converted by the fundamental curve unit that step 2) is extracted；
4) agglomerate layered clustering algorithm is utilized, cluster element is treated and is clustered, obtains battery under the corresponding operating condition of every kind of cluster
Operation curve.
2. method as described in claim 1, which is characterized in that the clustering factor X is electric current I, in voltage V, power P
Any one.
3. method as described in claim 1, which is characterized in that the specific method is as follows for the step 2):
The X-t curved section in operating condition to be analyzed between every two adjacent X zero point is extracted as the substantially bent of clustering factor X
Line unit；If not including the data point that X value is zero between adjacent in measurement data, X value opposite sign two data points, lead to
It crosses and two adjacent data point interpolations is handled, the zero point between supplement.
4. method as claimed in claim 3, which is characterized in that the specific method is as follows for the step 3):
Corresponding X is converted by each fundamental curve unit extracted in step 2)dY profile turns as element to be clustered
Change method are as follows:
For each fundamental curve unit,
Wherein, if X (t) > 0 in fundamental curve unit,
If X (t) < 0 in fundamental curve unit,
Wherein, X (t) represents the corresponding X value of t moment, t in fundamental curve unit X-t curved sectionsFor fundamental curve unit X-t curve
The start time point of section, teFor the termination time point of fundamental curve unit X-t curved section, g (Xd, t) and it is intermediate computations function, Xd
For operation variable.
5. method as described in claim 1, which is characterized in that specific step is as follows for the step 4):
4-1) it regard each element to be clustered as one kind independent；
4-2) calculate the class distance between any two class；The specific method is as follows:
For any two class, if class A hasTotal NAA element, class B haveTotal NBA element, then class A and class B
Between class distance D (A, B) be defined as the average distances of two class all elements, it may be assumed that
Wherein, d (li,l'j) it is element liWith element ljBetween element distance；Element distance is that two elements to be clustered are corresponding
XdThe gross area for the figure that Y profile and axis of abscissas and axis of ordinates surround；
4-3) using step 4-2's) as a result, this two class is merged into a new class apart from the smallest two class by selection class；
4-4) repeat step 4-2) to 4-3), until the sum of cluster reaches setting value N, cluster is finished, and obtains battery at every kind
The operation curve under corresponding operating condition is clustered, analysis finishes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910066652.8A CN109871881B (en) | 2019-01-24 | 2019-01-24 | Battery working condition analysis method based on agglomeration hierarchical clustering algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910066652.8A CN109871881B (en) | 2019-01-24 | 2019-01-24 | Battery working condition analysis method based on agglomeration hierarchical clustering algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109871881A true CN109871881A (en) | 2019-06-11 |
| CN109871881B CN109871881B (en) | 2020-12-08 |
Family
ID=66918003
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910066652.8A Active CN109871881B (en) | 2019-01-24 | 2019-01-24 | Battery working condition analysis method based on agglomeration hierarchical clustering algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109871881B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114463572A (en) * | 2022-03-01 | 2022-05-10 | 智慧足迹数据科技有限公司 | Region clustering method and related device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1160986A (en) * | 1995-11-30 | 1997-10-01 | 松下电器产业株式会社 | Component detecting method and apparatus |
| US20060221692A1 (en) * | 2005-04-05 | 2006-10-05 | Jian Chen | Compensating for coupling during read operations on non-volatile memory |
| CN103817091A (en) * | 2014-02-28 | 2014-05-28 | 清华大学 | Battery sorting method and system |
| CN103954913A (en) * | 2014-05-05 | 2014-07-30 | 哈尔滨工业大学深圳研究生院 | Predication method of electric vehicle power battery service life |
| CN108655028A (en) * | 2018-03-20 | 2018-10-16 | 中国电力科学研究院有限公司 | A kind of method and system classified to battery based on Fuzzy Mean Clustering Algorithm |
| US10163493B2 (en) * | 2017-05-08 | 2018-12-25 | International Business Machines Corporation | SRAM margin recovery during burn-in |
-
2019
- 2019-01-24 CN CN201910066652.8A patent/CN109871881B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1160986A (en) * | 1995-11-30 | 1997-10-01 | 松下电器产业株式会社 | Component detecting method and apparatus |
| US20060221692A1 (en) * | 2005-04-05 | 2006-10-05 | Jian Chen | Compensating for coupling during read operations on non-volatile memory |
| CN103817091A (en) * | 2014-02-28 | 2014-05-28 | 清华大学 | Battery sorting method and system |
| CN103954913A (en) * | 2014-05-05 | 2014-07-30 | 哈尔滨工业大学深圳研究生院 | Predication method of electric vehicle power battery service life |
| US10163493B2 (en) * | 2017-05-08 | 2018-12-25 | International Business Machines Corporation | SRAM margin recovery during burn-in |
| CN108655028A (en) * | 2018-03-20 | 2018-10-16 | 中国电力科学研究院有限公司 | A kind of method and system classified to battery based on Fuzzy Mean Clustering Algorithm |
Non-Patent Citations (3)
| Title |
|---|
| VENKATASAILANATHAN RAMADESIGAN ET,AL.: "Modeling and Simulation of Lithium-Ion Batteries from a Systems Engineering Perspective", 《JOURNAL OF THE ELECTROCHEMICAL SOCIETY》 * |
| 张剑波 等: "基于锂离子电池老化行为的析锂检测", 《万方全文数据库》 * |
| 郑路路 等: "基于上海市某型插电式混合动力汽车的行驶工况和电流工况研究", 《传动技术》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114463572A (en) * | 2022-03-01 | 2022-05-10 | 智慧足迹数据科技有限公司 | Region clustering method and related device |
| CN114463572B (en) * | 2022-03-01 | 2023-06-09 | 智慧足迹数据科技有限公司 | Regional clustering method and related device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109871881B (en) | 2020-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107662503B (en) | Electric automobile braking intention identification method based on acceleration and brake pedal states | |
| CN106208041B (en) | A kind of Distribution Network Harmonics current estimation method based on piecewise constant independent component analysis | |
| CN103645377B (en) | Based on the battery nominal power Forecasting Methodology of Dynamic array control algorithm | |
| CN105404350B (en) | A kind of thermoelectric power generation MPPT method based on two points of gradient method | |
| CN103793605A (en) | Lithium iron phosphate power battery equivalent circuit model parameter estimation method based on particle swarm algorithm | |
| CN109558897A (en) | Power grid planning multi-scene extraction method based on data mining | |
| CN101963640A (en) | Detection system and detection method of direct-current charging post | |
| CN104657520A (en) | Battery modeling method based on large capacity energy accumulation lithium ion battery | |
| CN105938578A (en) | Large-scale photovoltaic power station equivalent modeling method based on clustering analysis | |
| CN110031777A (en) | A method of quickly obtaining all single battery resistance values in battery pack | |
| CN112580289B (en) | Hybrid capacitor power state online estimation method and system | |
| CN106950856A (en) | MPPT modeling and simulating methods based on integrating mixed logic dynamic | |
| CN110470994A (en) | A kind of prediction technique of power battery peak power | |
| CN105868490A (en) | Multi-target selected harmonics suppression pulse width modulation method of modular multilevel converter | |
| CN109871881A (en) | A kind of battery performance analysis method based on agglomerate layered clustering algorithm | |
| CN111366864A (en) | Battery SOH on-line estimation method based on fixed voltage rise interval | |
| CN113627671A (en) | SOH prediction calculation method for single battery in energy storage scene | |
| Shuai et al. | Intelligent diagnosis of abnormal charging for electric bicycles based on improved dynamic time warping | |
| CN106707181A (en) | Cell parameter and charged state estimation method of lithium ion | |
| Feng et al. | Research on a battery test profile based on road test data from hybrid fuel cell buses | |
| CN113408617A (en) | XGboost and Stacking model fusion-based non-invasive load identification method | |
| CN106301056B (en) | Multiple-objection optimization SVPWM methods for MW class three-phase inversion system | |
| Wu et al. | Research on SOC hybrid estimation algorithm of power battery based on EKF | |
| Bao et al. | State of charge estimation for electric vehicle batteries based on LS-SVM | |
| Wang et al. | Identifying household EV models via weighted power recurrence graphs |
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 |