CN112307906A - Energy storage battery fault classification feature screening and dimension reduction method under neighbor propagation clustering - Google Patents
Energy storage battery fault classification feature screening and dimension reduction method under neighbor propagation clustering Download PDFInfo
- Publication number
- CN112307906A CN112307906A CN202011094254.6A CN202011094254A CN112307906A CN 112307906 A CN112307906 A CN 112307906A CN 202011094254 A CN202011094254 A CN 202011094254A CN 112307906 A CN112307906 A CN 112307906A
- Authority
- CN
- China
- Prior art keywords
- characteristic
- feature
- features
- energy storage
- matrix
- 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 57
- 238000004146 energy storage Methods 0.000 title claims abstract description 44
- 238000012216 screening Methods 0.000 title claims abstract description 15
- 239000011159 matrix material Substances 0.000 claims abstract description 55
- 238000007600 charging Methods 0.000 claims abstract description 7
- 238000007599 discharging Methods 0.000 claims abstract description 7
- 238000005065 mining Methods 0.000 claims abstract description 4
- 238000003745 diagnosis Methods 0.000 claims description 11
- 239000013598 vector Substances 0.000 claims description 9
- 238000013461 design Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 4
- 230000009286 beneficial effect Effects 0.000 claims description 3
- 238000013016 damping Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000011946 reduction process Methods 0.000 claims description 3
- 238000004880 explosion Methods 0.000 abstract description 3
- 230000002411 adverse Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 abstract description 2
- 238000001914 filtration Methods 0.000 description 5
- 238000013145 classification model Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000002349 favourable effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000004931 aggregating effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for screening and reducing dimension of fault classification characteristics of an energy storage battery under neighbor propagation clustering. The method comprises the following implementation processes: firstly, acquiring terminal voltage signal data samples of N energy storage batteries with different faults in the process of completing one-time charging and discharging, and mining the characteristics of the terminal voltage signals to form a characteristic set; secondly, defining a similarity matrix of the features by utilizing cosine similarity, clustering each feature based on a neighbor propagation method, and forming a plurality of feature clusters; calculating the mean value and the standard deviation of the mean value of each feature in the same cluster under different types of samples, and defining the feature with the maximum mean value standard deviation as a typical feature for representing the feature cluster; and taking the typical features of each feature cluster as the screened dimension-reduced features for the classification task. The method can automatically screen out the significant features for the classification task, relieve the problem of dimension explosion under the condition of small samples, reduce the adverse effect of redundant and invalid features on the performance of the classifier, and improve the accuracy of classification.
Description
The technical field is as follows:
the invention relates to the technical field of energy storage batteries, in particular to a method for screening and reducing dimension of fault classification features of an energy storage battery under neighbor propagation clustering.
Background art:
under the development of machine learning and artificial intelligence technology, the automation level of classification tasks of fault diagnosis of power equipment and an energy storage system is greatly improved. The feature engineering and the learner are key and core for solving the traditional classification task, and the significance of the mined features and the generalization capability of the learner are important conditions for restricting the accuracy, robustness and popularization capability of the classification task. At present, a great deal of research is available around the design and improvement of classification models, and in the face of different application requirements, there is no uniform and definite effective method for feature extraction. The method is often tentatively defined by indexes with physical meanings, and is easy to cause redundancy and invalid features to exist. Meanwhile, under the condition of limited data samples, the existence of invalid and redundant features causes the dimension of the feature space to be too high, the sample interval is too large, the dimension explosion phenomenon is formed, and the generalization capability of the classification model is reduced. Therefore, in the feature engineering, the extracted feature variables need to be selected or transformed to reduce the feature space dimension or enhance the significance of the features.
Currently, the characteristic selection modes of the existing energy storage system fault diagnosis can be roughly divided into a filtering type, a wrapping type and an embedding type. The filtering method is characterized in that feature selection is carried out according to the characteristics of a data set, then a learner is trained, namely the feature selection process is independent of the learner, the filtering method has the advantages that the calculated amount is small, the particularity of a learner model is not depended on, and the filtering method is difficult to carry out directional feature selection facing a classification task based on the characteristics of data; the wrapping method is characterized in that the performance of a used learning period is directly used as an evaluation criterion of the feature subset after feature selection, and has the advantages that the performance of a learner is used as an index to provide a direction for the feature selection process, but a learning machine needs to be trained for many times in the feature selection process, so that the calculation cost is high; the embedded method is characterized in that a characteristic selection process is related to a learner, the characteristic selection process is fused with a learner training process, and the characteristic selection is automatically carried out in the learner training process. Therefore, a new screening and dimension reduction method for fault diagnosis of the energy storage system is needed.
The invention content is as follows:
in order to improve the applicability of the energy storage system fault diagnosis feature selection method and reduce the feature selection calculation amount, the invention adopts a filtering type feature selection process, uses clustering thought for reference, realizes the distinguishing of feature sets on the basis of neighbor propagation clustering, analyzes the performance capability of each feature in the same type of feature clusters to different types of samples, and selects the typical feature with the maximized distinguishing capability. And the selection of the fault diagnosis characteristics of the energy storage system is completed, and simultaneously, the dimension reduction of the characteristic space is realized, and a foundation is laid for the design of a high-performance learning machine model.
The technical scheme adopted by the invention is as follows:
a method for screening and reducing dimension of fault classification features of energy storage batteries under neighbor propagation clustering comprises the following steps:
acquiring terminal voltage signal data samples of N energy storage batteries with different faults in the process of completing one-time charging and discharging, and mining the characteristics of the terminal voltage signals to form a characteristic set; the specific process is as follows:
step 1.1, acquiring N terminal voltage signal data samples of energy storage batteries with R faults in a one-time charging and discharging process by using an experimental measurement mode;
step 1.2, defining and normalizing the characteristic vectors of the energy storage end voltage characteristics represented by m characteristic variables according to a common characteristic extraction mode, namely the characteristic vector of the nth sample
Form feature set a { (X)(n),L(n))|n＝1,2,…,N；L(n)∈{L1,L2,…,LRAnd } expressed in matrix form as follows:
the rows represent samples, the top m columns represent features, and the m +1 columns represent categories of samples;
step 2, evaluating the similarity degree between the features by utilizing cosine similarity, and clustering the features by utilizing a neighbor propagation clustering method to form a plurality of feature clusters; the specific process is as follows:
step 2.1. forming a vector by using the values of all samples under the characteristic variables, namely
And i is 1,2, …, N, and calculating the cosine similarity sim between every two features, for example, the cosine similarity of the ith feature and the jth feature is:
forming a symmetrical square matrix with the number of the features as the number of rows (columns) according to the cosine similarity among the features, namely a similarity matrix S, wherein the negative value of the cosine similarity among the element features in the square matrix is as follows:
step 2.2, let t equal to 1, initialize the attraction matrix RetAnd the attribution degree matrix AvtAnd an attraction degree matrix RetAnd the attribution degree matrix AvtAre all in rows and columns equal to the similarity matrix SA square matrix, wherein a damping coefficient zeta and a maximum iteration number T are defined;
step 2.3, calculating and updating the attraction degree matrix Re of the t +1 generationt+1The (α, β) th element in the matrix is calculated as follows:
step 2.4, calculating and updating the attribution degree matrix Av of the t +1 generationt+1The (α, β) th element in the matrix is calculated as follows:
step 2.5, judging whether the maximum iteration times is reached, namely T is less than or equal to T; if yes, returning to the step 2.3 if t is equal to t +1, otherwise, entering the step 2.6;
step 2.6, let E ═ Ret+1+Avt+1Diagonalizing the matrix E, counting elements larger than zero on the diagonal line, and counting the characteristics of the row (column) numbers where the elements larger than zero are positioned, namely neighbor propagation clustering centers, selecting the columns where the clustering centers are positioned in the similarity matrix S, and selecting the category attributions of other characteristics according to the row selection to form a plurality of characteristic clusters; assuming that the eith and ej th diagonal elements in the diagonalized matrix E are greater than zero, the feature xeiAnd xejSelecting the eith column and the ej column in the similarity matrix S for the clustering center, and if-sim (ek, ei) is more than or equal to-sim (ek, ej) according to the sizes of-sim (ek, ei) and-sim (ek, ej), the ek th characteristic xekIs similar to the characteristic xeiBelongs to the feature xeiA category of (1); if-sim (ek, ei)<Sim (ek, ej), the ek-th feature xekSimilar toCharacteristic xejBelongs to the feature xejA category of (1); thus, two dry feature clusters can be formed for all the features, and the centers of the feature clusters are respectively formed by the feature xekAnd feature xejRepresents;
screening the characteristics which are most beneficial to fault diagnosis of the energy storage battery in the same characteristic cluster, and defining the characteristics as typical characteristics of the characteristic cluster; the method specifically comprises the following steps:
step 3.1, calculating the mean value of terminal voltage characteristic values of different fault energy storage battery samples under different similar characteristics in the same characteristic cluster according to the dividing result of the characteristic cluster obtained in the step 2; assume that in the kth feature cluster h (k) ═ xki,xkj,.. } characteristics xkiThe mean value of the characteristic values of the next different types of samples is muki＝[μki,1,μki,2,...,μki,R]Wherein the L isrClass sample is characterized by xkiThe mean of
Step 3.2, calculating the standard deviation of the terminal voltage characteristic value mean values of different fault energy storage battery samples under each characteristic in the same characteristic cluster in the step 3.1, and defining the characteristic with the maximum standard deviation (the difference of different fault energy storage batteries under the characteristic) in the characteristic cluster as the typical characteristic representing the characteristic cluster; assume that the kth feature cluster h (k) contains two feature variables, i.e., h (k) { x }ki,xkjFeature xkiAnd xkjMean value of terminal voltage characteristic values { mu ] of different fault energy storage battery sampleski,μkjThe standard deviations are respectively
Then selecting the characteristic with the largest standard deviation as the representative characteristic cluster H (k) typical characteristic xki(σki≥σkj)+xkj(σki<σkj)；
Step 4, forming a screened dimension reduction feature set according to the typical features of each feature cluster; and (3) removing other features of the atypical features in the step 1.2 according to the typical features of the feature clusters obtained in the step 3.2 to form a new feature set for describing voltage differences of the fault terminals of the R energy storage batteries, so as to realize the feature screening and dimension reduction process, be used for subsequent learning machine model design and finish the fault diagnosis task of the energy storage batteries.
Compared with the closest prior art, the method has the advantages that in the technical scheme, cosine similarity is used for describing the similarity degree of different features, a proximity propagation clustering method is further adopted, feature variables with high similarity degree are aggregated to form a feature cluster, then the mean values of the features in the same cluster in different types of samples are analyzed, the mean standard deviation is counted, and the feature with the largest mean standard deviation is selected as the typical feature representing the cluster. Compared with the existing multi-clustering method for sample clustering and other feature dimension reduction methods, the method provided by the invention has the advantages that the clustering process is used for aggregating feature variables to form a plurality of feature clusters for describing feature differences, and the typical features of each feature cluster are established in a most favorable classification mode, so that the feature screening process which is favorable for completing classification tasks is realized while the feature dimensions are greatly reduced, the problem of dimension explosion possibly caused by overhigh feature dimensions is reduced, the adverse effect of high feature dimensions on subsequent classification model design is alleviated, and the generalization capability and robustness of the classification tasks are favorably improved.
Description of the drawings:
FIG. 1 is a schematic flow diagram of the process of the present invention.
Fig. 2 is a schematic flow chart of forming a plurality of feature clusters by using a neighbor propagation clustering method in step 2.
The specific implementation mode is as follows:
example (b):
a method for screening and reducing dimension of fault classification features of energy storage batteries under neighbor propagation clustering comprises the following steps:
acquiring terminal voltage signal data samples of N energy storage batteries with different faults in the process of completing one-time charging and discharging, and mining the characteristics of the terminal voltage signals to form a characteristic set; the specific process is as follows:
step 1.1, acquiring N terminal voltage signal data samples of energy storage batteries with R faults in a one-time charging and discharging process by using an experimental measurement mode;
step 1.2, defining and normalizing the characteristic vectors of the energy storage end voltage characteristics represented by m characteristic variables according to a common characteristic extraction mode, namely the characteristic vector of the nth sample
Form feature set a { (X)(n),L(n))|n＝1,2,…,N；L(n)∈{L1,L2,…,LRAnd } expressed in matrix form as follows:
the rows represent samples, the top m columns represent features, and the m +1 columns represent categories of samples;
step 2, evaluating the similarity degree between the features by utilizing cosine similarity, and clustering the features by utilizing a neighbor propagation clustering method to form a plurality of feature clusters; the specific process is as follows:
step 2.1. forming a vector by using the values of all samples under the characteristic variables, namely
And i is 1,2, …, N, and calculating the cosine similarity sim between every two features, for example, the cosine similarity of the ith feature and the jth feature is:
forming a symmetrical square matrix with the number of the features as the number of rows (columns) according to the cosine similarity among the features, namely a similarity matrix S, wherein the negative value of the cosine similarity among the element features in the square matrix is as follows:
step 2.2, let t equal to 1, initialize the attraction matrix RetAnd the attribution degree matrix AvtAnd an attraction degree matrix RetAnd the attribution degree matrix AvtThe method comprises the following steps that (1) a similarity matrix S is an equal row and column square matrix, and a damping coefficient zeta and a maximum iteration time T are defined;
step 2.3, calculating and updating the attraction degree matrix Re of the t +1 generationt+1The (α, β) th element in the matrix is calculated as follows:
step 2.4, calculating and updating the attribution degree matrix Av of the t +1 generationt+1The (α, β) th element in the matrix is calculated as follows:
step 2.5, judging whether the maximum iteration times is reached, namely T is less than or equal to T; if yes, returning to the step 2.3 if t is equal to t +1, otherwise, entering the step 2.6;
step 2.6, let E ═ Ret+1+Avt+1Diagonalizing the matrix E, counting elements larger than zero on the diagonal line, and counting the characteristics of the row (column) numbers where the elements larger than zero are positioned, namely neighbor propagation clustering centers, selecting the columns where the clustering centers are positioned in the similarity matrix S, and selecting the category attributions of other characteristics according to the row selection to form a plurality of characteristic clusters; diagonal in the hypothesis diagonalization matrix EElement ei and ej are greater than zero, then feature xeiAnd xejSelecting the eith column and the ej column in the similarity matrix S for the clustering center, and if-sim (ek, ei) is more than or equal to-sim (ek, ej) according to the sizes of-sim (ek, ei) and-sim (ek, ej), the ek th characteristic xekIs similar to the characteristic xeiBelongs to the feature xeiA category of (1); if-sim (ek, ei)<Sim (ek, ej), the ek-th feature xekIs similar to the characteristic xejBelongs to the feature xejA category of (1); thus, two dry feature clusters can be formed for all the features, and the centers of the feature clusters are respectively formed by the feature xekAnd feature xejRepresents;
screening the characteristics which are most beneficial to fault diagnosis of the energy storage battery in the same characteristic cluster, and defining the characteristics as typical characteristics of the characteristic cluster; the method specifically comprises the following steps:
step 3.1, calculating the mean value of terminal voltage characteristic values of different fault energy storage battery samples under different similar characteristics in the same characteristic cluster according to the dividing result of the characteristic cluster obtained in the step 2; assume that in the kth feature cluster h (k) ═ xki,xkj,.. } characteristics xkiMean of the following different classes of sample eigenvalues:
μki＝[μki,1,μki,2,...,μki,R]wherein the L isrClass sample is characterized by xkiThe mean of
Step 3.2, calculating the standard deviation of the terminal voltage characteristic value mean values of different fault energy storage battery samples under each characteristic in the same characteristic cluster in the step 3.1, and defining the characteristic with the maximum standard deviation (the difference of different fault energy storage batteries under the characteristic) in the characteristic cluster as the typical characteristic representing the characteristic cluster; assume that the kth feature cluster h (k) contains two feature variables, i.e., h (k) { x }ki,xkjFeature xkiAnd xkjMean value of terminal voltage characteristic values { mu ] of different fault energy storage battery sampleski,μkjThe standard deviations are respectively
And
then selecting the characteristic with the largest standard deviation as the representative characteristic cluster H (k) typical characteristic xki(σki≥σkj)+xkj(σki<σkj)；
Step 4, forming a screened dimension reduction feature set according to the typical features of each feature cluster; and (3) removing other features of the atypical features in the step 1.2 according to the typical features of the feature clusters obtained in the step 3.2 to form a new feature set for describing voltage differences of the fault terminals of the R energy storage batteries, so as to realize the feature screening and dimension reduction process, be used for subsequent learning machine model design and finish the fault diagnosis task of the energy storage batteries.
Finally, it should be noted that the described embodiments are only some, but not all embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Claims (1)
1. A method for screening and reducing dimension of fault classification features of energy storage batteries under neighbor propagation clustering is characterized by comprising the following steps:
acquiring terminal voltage signal data samples of N energy storage batteries with different faults in the process of completing one-time charging and discharging, and mining the characteristics of the terminal voltage signals to form a characteristic set; the specific process is as follows:
step 1.1, acquiring N terminal voltage signal data samples of energy storage batteries with R faults in a one-time charging and discharging process by using an experimental measurement mode;
step 1.2, defining and normalizing the characteristic vectors of the energy storage end voltage characteristics represented by m characteristic variables according to a common characteristic extraction mode, namely the characteristic vector of the nth sample
Form feature set a { (X)(n),L(n))|n＝1,2,…,N；L(n)∈{L1,L2,…,LRAnd } expressed in matrix form as follows:
the rows represent samples, the top m columns represent features, and the m +1 columns represent categories of samples;
step 2, evaluating the similarity degree between the features by utilizing cosine similarity, and clustering the features by utilizing a neighbor propagation clustering method to form a plurality of feature clusters; the specific process is as follows:
step 2.1. forming a vector by using the values of all samples under the characteristic variables, namely
Calculating the cosine similarity sim between every two characteristics, wherein the cosine similarity of the ith characteristic and the jth characteristic is as follows:
forming a symmetrical square matrix with the number of the features as the number of rows (columns) according to the cosine similarity among the features, namely a similarity matrix S, wherein the negative value of the cosine similarity among the element features in the square matrix is as follows:
step 2.2, let t equal to 1, initialize the attraction matrix RetAnd the attribution degree matrix AvtAnd an attraction degree matrix RetAnd the attribution degree matrix AvtThe method comprises the following steps that (1) a similarity matrix S is an equal row and column square matrix, and a damping coefficient zeta and a maximum iteration time T are defined;
step 2.3, calculating and updating the attraction degree matrix Re of the t +1 generationt+1The (α, β) th element in the matrix is calculated as follows:
step 2.4, calculating and updating the attribution degree matrix Av of the t +1 generationt+1The (α, β) th element in the matrix is calculated as follows:
step 2.5, judging whether the maximum iteration times is reached, namely T is less than or equal to T; if yes, returning to the step 2.3 if t is equal to t +1, otherwise, entering the step 2.6;
step 2.6, let E ═ Ret+1+Avt+1Diagonalizing the matrix E, counting elements larger than zero on the diagonal line, and counting the characteristics of the row (column) numbers where the elements larger than zero are positioned, namely neighbor propagation clustering centers, selecting the columns where the clustering centers are positioned in the similarity matrix S, and selecting the category attributions of other characteristics according to the row selection to form a plurality of characteristic clusters; assuming that the eith and ej th diagonal elements in the diagonalized matrix E are greater than zero, the feature xeiAnd xejSelecting the eith column and the ej column in the similarity matrix S for the clustering center, and if-sim (ek, ei) is more than or equal to-sim (ek, ej) according to the sizes of-sim (ek, ei) and-sim (ek, ej), the ek th characteristic xekIs similar to the characteristic xeiBelongs to the feature xeiA category of (1); if-sim (ek, ei)<Sim (ek, ej), the ek-th feature xekIs similar to the characteristic xejBelongs to the feature xejA category of (1); thus, two dry feature clusters can be formed for all the features, and the centers of the feature clusters are respectively formed by the feature xekAnd feature xejRepresents;
screening the characteristics which are most beneficial to fault diagnosis of the energy storage battery in the same characteristic cluster, and defining the characteristics as typical characteristics of the characteristic cluster; the method specifically comprises the following steps:
step 3.1, calculating the mean value of terminal voltage characteristic values of different fault energy storage battery samples under different similar characteristics in the same characteristic cluster according to the dividing result of the characteristic cluster obtained in the step 2; assume that in the kth feature cluster h (k) ═ xki,xkj,.. } characteristics xkiThe mean value of the characteristic values of the next different types of samples is muki＝[μki,1,μki,2,...,μki,R]Wherein the L isrClass sample is characterized by xkiThe mean of
Step 3.2, calculating the standard deviation of the terminal voltage characteristic value mean values of different fault energy storage battery samples under each characteristic in the same characteristic cluster in the step 3.1, and defining the characteristic with the maximum standard deviation (the difference of different fault energy storage batteries under the characteristic) in the characteristic cluster as the typical characteristic representing the characteristic cluster; assume that the kth feature cluster h (k) contains two feature variables, i.e., h (k) { x }ki,xkjFeature xkiAnd xkjMean value of terminal voltage characteristic values { mu ] of different fault energy storage battery sampleski,μkjThe standard deviations are respectively
And
then selecting the characteristic with the largest standard deviation as the representative characteristic cluster H (k) typical characteristic xki(σki≥σkj)+xkj(σki<σkj)；
Step 4, forming a screened dimension reduction feature set according to the typical features of each feature cluster;
and (3) removing other features of the atypical features in the step 1.2 according to the typical features of the feature clusters obtained in the step 3.2 to form a new feature set for describing voltage differences of the fault terminals of the R energy storage batteries, so as to realize the feature screening and dimension reduction process, be used for subsequent learning machine model design and finish the fault diagnosis task of the energy storage batteries.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011094254.6A CN112307906B (en) | 2020-10-14 | 2020-10-14 | Energy storage battery fault classification feature screening dimension reduction method under neighbor propagation clustering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011094254.6A CN112307906B (en) | 2020-10-14 | 2020-10-14 | Energy storage battery fault classification feature screening dimension reduction method under neighbor propagation clustering |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112307906A true CN112307906A (en) | 2021-02-02 |
| CN112307906B CN112307906B (en) | 2023-07-04 |
Family
ID=74488060
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011094254.6A Active CN112307906B (en) | 2020-10-14 | 2020-10-14 | Energy storage battery fault classification feature screening dimension reduction method under neighbor propagation clustering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112307906B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113552496A (en) * | 2021-06-29 | 2021-10-26 | 哈尔滨理工大学 | Voltage cosine similarity-based diagnosis method for short circuit fault in battery series module |
| CN114088388A (en) * | 2021-12-10 | 2022-02-25 | 华润电力技术研究院有限公司 | Fault diagnosis method and fault diagnosis device for gearbox |
| CN114497770A (en) * | 2022-01-26 | 2022-05-13 | 上海玫克生智能科技有限公司 | Method, system and terminal for analyzing state of battery box in battery cluster |
| CN114818881A (en) * | 2022-04-07 | 2022-07-29 | 青岛大学 | Fault detection and positioning method for voltage sensor of vehicle-mounted power battery pack |
| CN117457094A (en) * | 2023-12-20 | 2024-01-26 | 安徽农业大学 | Oxyhydrogen fuel cell energy efficiency performance evaluation method and system based on AP algorithm |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010064414A1 (en) * | 2008-12-02 | 2010-06-10 | ソニー株式会社 | Gene clustering program, gene clustering method, and gene cluster analyzing device |
| JP2016197406A (en) * | 2015-04-06 | 2016-11-24 | 国立研究開発法人産業技術総合研究所 | Information processor, information processing system, information processing method, program, and recording medium |
| CN109245100A (en) * | 2018-11-07 | 2019-01-18 | 国网浙江省电力有限公司经济技术研究院 | Consider the Dynamic Load Modeling method of alternating current-direct current distribution network load composition time variation |
| CN109976308A (en) * | 2019-03-29 | 2019-07-05 | 南昌航空大学 | A kind of extracting method of the fault signature based on Laplce's score value and AP cluster |
| CN110084520A (en) * | 2019-04-30 | 2019-08-02 | 国网上海市电力公司 | Charging station site selecting method and device based on public bus network Yu gridding AP algorithm |
| WO2020066257A1 (en) * | 2018-09-26 | 2020-04-02 | 国立研究開発法人理化学研究所 | Classification device, classification method, program, and information recording medium |
| CN111506730A (en) * | 2020-04-17 | 2020-08-07 | 腾讯科技（深圳）有限公司 | Data clustering method and related device |
-
2020
- 2020-10-14 CN CN202011094254.6A patent/CN112307906B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010064414A1 (en) * | 2008-12-02 | 2010-06-10 | ソニー株式会社 | Gene clustering program, gene clustering method, and gene cluster analyzing device |
| JP2016197406A (en) * | 2015-04-06 | 2016-11-24 | 国立研究開発法人産業技術総合研究所 | Information processor, information processing system, information processing method, program, and recording medium |
| WO2020066257A1 (en) * | 2018-09-26 | 2020-04-02 | 国立研究開発法人理化学研究所 | Classification device, classification method, program, and information recording medium |
| CN109245100A (en) * | 2018-11-07 | 2019-01-18 | 国网浙江省电力有限公司经济技术研究院 | Consider the Dynamic Load Modeling method of alternating current-direct current distribution network load composition time variation |
| CN109976308A (en) * | 2019-03-29 | 2019-07-05 | 南昌航空大学 | A kind of extracting method of the fault signature based on Laplce's score value and AP cluster |
| CN110084520A (en) * | 2019-04-30 | 2019-08-02 | 国网上海市电力公司 | Charging station site selecting method and device based on public bus network Yu gridding AP algorithm |
| CN111506730A (en) * | 2020-04-17 | 2020-08-07 | 腾讯科技（深圳）有限公司 | Data clustering method and related device |
Non-Patent Citations (2)
| Title |
|---|
| ZHIWEI HE等: "Battery Grouping with Time Series Clustering Based on Affinity Propagation", 《ENERGIES 》 * |
| 唐丽君: "基于近邻传播算法的电池配组技术研究", 《中国优秀硕士学位论文全文数据库 (工程科技Ⅱ辑)》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113552496A (en) * | 2021-06-29 | 2021-10-26 | 哈尔滨理工大学 | Voltage cosine similarity-based diagnosis method for short circuit fault in battery series module |
| CN113552496B (en) * | 2021-06-29 | 2024-04-02 | 哈尔滨理工大学 | Battery series module internal short circuit fault diagnosis method based on voltage cosine similarity |
| CN114088388A (en) * | 2021-12-10 | 2022-02-25 | 华润电力技术研究院有限公司 | Fault diagnosis method and fault diagnosis device for gearbox |
| CN114497770A (en) * | 2022-01-26 | 2022-05-13 | 上海玫克生智能科技有限公司 | Method, system and terminal for analyzing state of battery box in battery cluster |
| CN114818881A (en) * | 2022-04-07 | 2022-07-29 | 青岛大学 | Fault detection and positioning method for voltage sensor of vehicle-mounted power battery pack |
| CN114818881B (en) * | 2022-04-07 | 2024-04-26 | 青岛大学 | Fault detection and positioning method for voltage sensor of vehicle-mounted power battery pack |
| CN117457094A (en) * | 2023-12-20 | 2024-01-26 | 安徽农业大学 | Oxyhydrogen fuel cell energy efficiency performance evaluation method and system based on AP algorithm |
| CN117457094B (en) * | 2023-12-20 | 2024-03-29 | 安徽农业大学 | Oxyhydrogen fuel cell energy efficiency performance evaluation method and system based on AP algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112307906B (en) | 2023-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112307906B (en) | Energy storage battery fault classification feature screening dimension reduction method under neighbor propagation clustering | |
| CN110752410B (en) | Method for rapidly sorting and recombining retired lithium battery | |
| CN112379269B (en) | Battery abnormality detection model training and detection method and device thereof | |
| CN109657584B (en) | Improved LeNet-5 fusion network traffic sign identification method for assisting driving | |
| CN110516305B (en) | Intelligent fault diagnosis method under small sample based on attention mechanism meta-learning model | |
| CN109635928A (en) | A kind of voltage sag reason recognition methods based on deep learning Model Fusion | |
| CN110726898B (en) | Power distribution network fault type identification method | |
| CN115277888B (en) | Method and system for analyzing message type of mobile application encryption protocol | |
| CN112116950B (en) | Protein folding identification method based on depth measurement learning | |
| CN111582372B (en) | Image classification method, model, storage medium and electronic device | |
| CN113033678A (en) | Lithium battery pack fault diagnosis method based on adaptive countermeasure network | |
| CN115659254A (en) | Power quality disturbance analysis method for power distribution network with bimodal feature fusion | |
| CN116307059A (en) | Power distribution network region fault prediction model construction method and device and electronic equipment | |
| CN112287980A (en) | Power battery screening method based on typical feature vector | |
| CN115327041A (en) | Air pollutant concentration prediction method based on correlation analysis | |
| CN116484299A (en) | Charging pile fault diagnosis method based on integration of gradient lifting tree and multi-layer perceptron | |
| CN114898136A (en) | Small sample image classification method based on feature self-adaption | |
| CN111783688B (en) | Remote sensing image scene classification method based on convolutional neural network | |
| CN110261773B (en) | Aviation generator fault symptom extraction method and system | |
| CN115761654B (en) | Vehicle re-identification method | |
| CN110943473A (en) | Generator coherence identification method based on wide area measurement system and clustering theory | |
| CN102609732A (en) | Object recognition method based on generalization visual dictionary diagram | |
| CN113361631A (en) | Insulator aging spectrum classification method based on transfer learning | |
| CN114997214A (en) | Fault diagnosis method and device for residual error intensive network | |
| CN113505817A (en) | Self-adaptive weighting training method for bearing fault diagnosis model samples under unbalanced data |
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 |