CN107656898B - Demand response resource clustering method - Google Patents
Demand response resource clustering method Download PDFInfo
- Publication number
- CN107656898B CN107656898B CN201710958627.1A CN201710958627A CN107656898B CN 107656898 B CN107656898 B CN 107656898B CN 201710958627 A CN201710958627 A CN 201710958627A CN 107656898 B CN107656898 B CN 107656898B
- Authority
- CN
- China
- Prior art keywords
- resource
- resources
- clustering
- representing
- demand response
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/141—Discrete Fourier transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Discrete Mathematics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a demand response resource clustering method, which comprises the following steps: the initial operation comprises the following steps: step 1, performing discrete Fourier transform on loads of all resources in a park in the previous day, and extracting the first five Fourier components; step 2, standardizing the extracted components and the characteristic quantity of the resources to form a standard characteristic vector of the resources in a single day; step 3, setting a clustering center and a clustering end condition, and clustering resources based on an FCM algorithm and standard feature vectors of resource single day until the clustering end condition is met; and 4, updating the clusters according to the clustering result in the step 3. The traditional resource classification mode according to information of workers, businessmen, residents and the like is broken through, and clustering of response resources is achieved, so that the efficiency of demand response is remarkably improved, and the response standard reaching rate of the resources is improved.
Description
Technical Field
The invention relates to a demand response resource clustering method.
Background
Under the background of power reform, the smart grid faces various different service scenes such as new energy consumption on site, rapid emergency response, differentiated customer service and the like. The service scenes prompt the power grid company to classify the resources when executing the demand response, so that the service is pertinently provided, and the demand response efficiency is improved.
Conventional demand response classifications are often classified by industry category or load size, such as industrial users, commercial users, residential users, and the like. However, such classification is rough, industrial users can be subdivided into many different industries, commercial users have different business hours and characteristics, and residential users also have differences in work and care. This classification increases the work difficulty for the system administrator and the instability of the system.
Existing clustering algorithms can solve the above problem to some extent, but these clustering algorithms generally rely on demand response events that have already been performed. However, for most systems, the number of demand responses per year is not large, and the clustering effect shows great randomness due to the limited sample size.
Disclosure of Invention
Aiming at the problem that the resource execution rate is low when the system executes the demand response, the characteristic components are extracted according to the load curve and the characteristics of the resource for clustering. The traditional resource classification mode according to information of workers, businessmen, residents and the like is broken through, and clustering of response resources is achieved, so that the efficiency of demand response is remarkably improved, and the response standard reaching rate of the resources is improved.
In order to achieve the technical purpose and achieve the technical effect, the invention is realized by the following technical scheme:
a demand response resource clustering method comprises the following steps during initial operation:
step 2, standardizing the extracted components and the characteristic quantity of the resources to form a standard characteristic vector of the resources in a single day;
and 4, updating the clusters according to the clustering result in the step 3.
Preferably, when the system is not operated for the first time, the membership degree of each resource is updated according to the aging function, and the clusters are corrected according to the regulation and control effect of demand response.
Preferably, in step 1, the first five components are extracted according to the following formula:
wherein p (N) is the load value of the resource at the time point N, and N is the sampling times of the resource in one day.
Preferably, step 2 comprises the steps of:
step 201, using Fa1、Fb1、Fa2、Fb2Each representing a component after unitization, then:
step 202, the original characteristic components of the resource in the previous day are as follows:
(Cdescend,CLifting of wine,COutput force,SResponse to,TNotification,Fa1,Fb1,Fa2,Fb2)
In the formula, CDescend、CLifting of wine、COutput forceRespectively representing the load reduction capacity, the load increase capacity and the output capacity of the resource; sResponse to、TNotificationRespectively representing the speed of resource response and the duration of advance notification;
step 203, standardizing the original characteristic components of the resource in the previous day, wherein:
in the formula, σDescendC representing all resourcesDescendCorresponding variance, σLifting of wineC representing all resourcesLifting of wineCorresponding variance, σOutput forceC representing all resourcesOutput forceCorresponding variance, σResponse toS representing all resourcesResponse toCorresponding variance, σNotificationT representing all resourcesNotificationCorresponding variance, σa1F representing all resourcesa1Corresponding variance, σa2F representing all resourcesa2Corresponding variance, σb1F representing all resourcesb1Corresponding variance, σb2F representing all resourcesb2The corresponding variance;
step 204, forming a standard feature vector of the resource for a single day as:
preferably, step 3 specifically comprises the following steps:
step 301, user sets the number of categories c, each category psiiInitial cluster center ciMaximum number of cycles smaxAnd initial degree of membership muij′;
Step 302 of utilizing the clustering centers c obtained in step 301i' calculation of degree of membership muij', wherein the degree of membership μijIs resource thetajBelonging to the classification psiiDegree of (c), let resource θjThe standard feature vector of a single day is xj:
In the formula, | | · | |, represents the distance of two feature vectors;
step 303, obtaining the membership degree mu according to the step 302ijRecalculating cluster centers ci′:
Step 304, settingRepresenting the resource theta obtained by the s-th calculationjBelonging to the classification psiiDegree of membership of, calculatingOr s is greater than or equal to smaxWhether or not it holds, where is the algorithm precision, smaxIs the last round;
step 305, if aboves≥smaxIf the two formulas are not satisfied, the step 302 is skipped to continue the circulation; if one formula is established, the loop ends and the resource thetajAmong the corresponding membership degrees, the membership degree with the maximum value is the resource thetajAnd (4) corresponding classification.
Preferably, the data is updated every day according to the load data of the previous dayThe cluster center before the update is represented,and representing the updated cluster center and membership degree, and then updating the algorithm as follows:
preferably, when the system is not operated for the first time, the membership degree of the resource after the demand response is executed is corrected, if the system only marks whether the resource reaches the standard, and if the user executes the demand response to several types in the existing classifications, then:
step 401, if the resource is theta'jIf the aging function is not met, the aging function f (t) is:
f(t)=1-e-t
wherein t represents the number of days until the demand response event;
step 402, updating the membership degree of the resource according to the following formula:
wherein the content of the first and second substances,represents the corrected membership degree, c'1、c'2…c'dIndicating the cluster centers to which the classification of the demand response was performed.
Preferably, when the system is not operated for the first time, the membership degree of the resource after the demand response is executed is corrected, and if the system expresses the standard reaching condition of the resource in a scoring mode, the system comprises the following steps:
step 501, correcting coefficient f'j(t) the following:
wherein s isjThe score of the corresponding resource is sigma which is the fraction of the resource reaching the standard; f (t) is an aging function;
step 502, setting a decision formula delta, and calculating according to the following formula:
step 503, if Δ is greater than or equal to 1, the correction formula of the membership degree is as follows:
if Delta is less than 1, the correction membership is calculated according to the following formula:
representing the corrected membership degree; c'1、c'2…c'dIndicating the cluster centers to which the classification of the demand response was performed.
The invention has the beneficial effects that:
firstly, the load curve of the user is closely related to the regulation and control capability of the user on the load, the load curve characteristics are extracted and used as the basis of clustering, and the influence caused by insufficient samples can be made up. Therefore, combining the load curve, the resource characteristics and the performed demand response is an effective idea for solving the problem of demand response resource clustering.
Secondly, the demand response resource clustering algorithm provided by the invention can effectively improve the operation speed of the algorithm by extracting the Fourier coefficient from the load curve, and simultaneously, the characteristics of the load curve are brought into the influence factors of resource clustering.
Thirdly, the demand response resource clustering algorithm provided by the invention uniformly analyzes the characteristics of the resources and the influence of the historical demand response effect on the resources through the setting of the aging function. The method can better fit with actual requirements, and therefore higher resource response rate is obtained.
Drawings
FIG. 1 is a flow chart of a method for primary clustering of demand response resources according to the present invention;
FIG. 2 is a flow chart of a non-primary clustering method for demand response resources according to the present invention;
fig. 3 is a comparison diagram before and after feature component extraction of resources in the embodiment of the present invention.
Detailed Description
The present invention will be better understood and implemented by those skilled in the art by the following detailed description of the technical solution of the present invention with reference to the accompanying drawings and specific examples, which are not intended to limit the present invention.
As shown in fig. 1, the initial operation of a demand response resource clustering method includes the following steps:
step 2, standardizing the extracted components and other characteristic quantities of the resources to form standard characteristic vectors of the resources in a single day;
and 3, setting a clustering center and a clustering ending condition, and clustering the resources based on the FCM algorithm and the standard characteristic vector of the resource in a single day until the clustering ending condition is met. The method is based on an FCM (Fuzzy C-means) algorithm and a standard feature vector of a resource single day, and used for clustering resources;
and 4, updating the clusters every day according to the clustering result in the step 3.
Preferably, when the system is not operated for the first time, the membership degree of each resource is updated according to the aging function, and the clusters are corrected according to the regulation and control effect of demand response.
The invention provides a demand response resource clustering algorithm, and belongs to the technical field of demand response of power systems. The algorithm firstly utilizes Fourier series to select five-dimensional low-frequency components, and compression and extraction of load information are realized. And then representing the demand response resources in a vector form by combining the output characteristics, the response characteristics and the like of the resources. Aiming at the problem that the variances of the components are different, a representation space of the response resources is constructed through vector unitization, and the distance is defined under the space. Subsequently, in a demand response scene, a resource clustering algorithm based on the fuzzy C mean is constructed. And finally, fully mining the executed demand response effect of the system by the algorithm, and realizing the correction of the resource clustering by using the aging function and the grading of the response standard-reaching condition aiming at two different standard-reaching condition representation modes. The invention can break through the traditional resource classification mode according to information of workers, businessmen, residents and the like, and realize the clustering of response resources, thereby obviously improving the efficiency of demand response and improving the response standard reaching rate of the resources, and the detailed description is given below.
Preferably, in step 1, the first five components are extracted according to the following formula:
wherein p (N) is the load value of the resource at the time point N, and N is the sampling times of the resource in one day.
Preferably, step 2 comprises the steps of:
step 201, using Fa1、Fb1、Fa2、Fb2Each representing a component after unitization, then:
step 202, the original characteristic components of the resource in the previous day are as follows:
(Cdescend,CLifting of wine,COutput force,SResponse to,TNotification,Fa1,Fb1,Fa2,Fb2)
In the formula, CDescend、CLifting of wine、COutput forceRespectively representing the load reduction capacity, the load increase capacity and the output capacity of the resource; sResponse to、TNotificationRespectively representing the speed of resource response and the duration of advance notification;
step 203, standardizing the original characteristic components of the resource in the previous day, wherein:
in the formula, σDescendC representing all resourcesDescendCorresponding variance, σLifting of wineC representing all resourcesLifting of wineCorresponding variance, σOutput forceC representing all resourcesOutput forceCorresponding variance, σResponse toS representing all resourcesResponse toCorresponding variance, σNotificationT representing all resourcesNotificationCorresponding variance, σa1F representing all resourcesa1Corresponding variance, σa2F representing all resourcesa2Corresponding variance, σb1F representing all resourcesb1Corresponding variance, σb2F representing all resourcesb2The corresponding variance;
step 204, forming a standard feature vector of the resource for a single day as:
preferably, step 3 specifically comprises the following steps:
step 301, user sets the number of categories c, each category psiiInitial cluster center ciMaximum number of cycles smaxAnd initial degree of membership muij;
Step 302 of utilizing the clustering centers c obtained in step 301i' calculation of degree of membership muij', wherein the degree of membership μijIs resource thetajBelonging to the classification psiiDegree of (c), let resource θjThe standard feature vector of a single day is xj:
In the formula, | | |, represents the distance between two feature vectors, expressed as | | | xj-ciFor example, | |:
step 303, obtaining the membership degree mu according to the step 302ijRecalculating cluster centers ci′:
Step 304, settingRepresenting the resource theta obtained by the s-th calculationjBelonging to the classification psiiDegree of membership of, calculatingOr s is greater than or equal to smaxWhether the algorithm is established or not is the algorithm precision, and the algorithm precision can be set according to actual requirements; smaxIs the last round;
step 305, if aboves≥smaxIf the two formulas are not satisfied, the step 302 is skipped to continue the circulation; if one formula is established, the loop ends and the resource thetajAmong the corresponding membership degrees, the membership degree with the maximum value is the resource thetajAnd (4) corresponding classification.
Preferably, in step 4, the load data is updated every day according to the load data of the previous dayThe cluster center before the update is represented,and representing the updated cluster center and membership degree, and then updating the algorithm as follows:
preferably, as shown in fig. 2, when the system is not operated for the first time, the membership degree of the resource after the execution of the demand response is modified, and if the system only marks whether the resource reaches the standard (0-1 way), and if the user executes the demand response to several categories among the existing categories, then:
step 401, if the resource is theta'jIf the aging function is not met, the aging function f (t) is:
f(t)=1-e-t
wherein t represents the number of days until the demand response event;
step 402, updating the membership degree of the resource according to the following formula:
wherein the content of the first and second substances,represents the corrected membership degree, c'1、c'2…c'dIndicating the cluster centers to which the classification of the demand response was performed.
If the system expresses the attainment of the resource in a scoring manner (a point-valued manner), the correction factor is as follows:
step 501, correcting coefficient f'j(t) the following:
wherein s isjThe score of the corresponding resource is sigma which is the fraction of the resource reaching the standard; f (t) is an aging function;
step 502, setting a decision formula delta, and calculating according to the following formula:
step 503, if Δ is greater than or equal to 1, the correction formula of the membership degree is as follows:
if Delta is less than 1, the correction membership is calculated according to the following formula:
represents the corrected membership degree, c'1、c’2…c’dIndicating the cluster centers to which the classification of the demand response was performed.
218 demand response resources are selected, the categories of residential users, hotels, business halls, shopping malls and the like are covered, characteristic components are extracted from the single-day loads of the resources, the demand response resources are clustered by adopting the method, and fig. 3 is a comparison graph before and after the characteristic components of the resources are extracted.
Standardizing the characteristic components of the resources to obtain standardized characteristic components of each resource:
all resources are classified into six categories according to a plan, including multi-period easily-regulated users (type 1 resources), early peak easily-regulated users (type 2 resources), mid-peak easily-regulated users (type 3 resources), late peak easily-regulated users (type 4 resources), inert users (type 5 resources) and energy storage devices (type 6 resources).
By utilizing the FCM algorithm, 51 types of resources are obtained, 23 types of resources are obtained, 83 types of resources are obtained, 42 types of resources are obtained, 14 types of resources are obtained, and 5 types of resources are obtained. After that, the classification of the users is regularly corrected according to the daily load curve of the resource and the actual demand response condition.
The demand response resource clustering algorithm comprehensively considers the actual conditions of less demand response times and less samples, combines the load curve of the resource, realizes the clustering of the demand response resources, and is more accurate compared with the existing method; the invention can be popularized to solve other types of intelligent power grid parks and also can be popularized to industrial parks containing intelligent equipment, and the expansion of demand response business is facilitated.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (2)
1. A demand response resource clustering method is characterized by comprising the following steps in the initial operation:
step 1, performing discrete Fourier transform on loads of all resources in a park in the previous day, and extracting the first five Fourier components;
step 2, standardizing the extracted components and the characteristic quantity of the resources to form a standard characteristic vector of the resources in a single day;
step 3, setting a clustering center and a clustering end condition, and clustering resources based on an FCM algorithm and standard feature vectors of resource single day until the clustering end condition is met;
step 4, updating the clusters according to the clustering result in the step 3;
when the system is not in initial operation, updating the membership degree of each resource according to the aging function, and correcting the clusters according to the regulation and control effect of demand response;
in step 1, the first five components are extracted according to the following formula:
wherein, p (N) is the load value of the resource at the time point N, and N is the sampling times of the resource in one day;
the step 2 comprises the following steps:
step 201, using Fa1、Fb1、Fa2、Fb2Each representing a component after unitization, then:
step 202, the original characteristic components of the resource in the previous day are as follows:
(Cdescend,CLifting of wine,COutput force,SResponse to,TNotification,Fa1,Fb1,Fa2,Fb2)
In the formula, CDescend、CLifting of wine、COutput forceRespectively representing the load reduction capacity, the load increase capacity and the output capacity of the resource; sResponse to、TNotificationRespectively representing the speed of resource response and the duration of advance notification;
step 203, standardizing the original characteristic components of the resource in the previous day, wherein:
in the formula, σDescendC representing all resourcesDescendCorresponding variance, σLifting of wineC representing all resourcesLifting of wineCorresponding variance, σOutput forceC representing all resourcesOutput forceCorresponding variance, σResponse toS representing all resourcesResponse toCorresponding variance, σNotificationT representing all resourcesNotificationCorresponding variance, σa1F representing all resourcesa1Corresponding variance, σa2F representing all resourcesa2Corresponding variance, σb1F representing all resourcesb1Corresponding variance, σb2F representing all resourcesb2The corresponding variance;
step 204, forming a standard feature vector of the resource for a single day as:
the step 3 specifically comprises the following steps:
step 301, user sets the number of categories c, each category psiiInitial cluster center ciMaximum number of cycles smaxAnd initial degree of membership muij;
Step 302 of utilizing the clustering centers c obtained in step 301i' calculation of degree of membership muij', wherein the degree of membership μijIs resource thetajBelonging to the classification psiiDegree of (c), let resource θjThe standard feature vector of a single day is xj:
In the formula, | | · | |, represents the distance of two feature vectors;
step 303, obtaining the membership degree mu according to the step 302ij' recalculation of respective clustering centers ci′:
Step 304, settingRepresenting the resource theta obtained by the s-th calculationjBelonging to the classification psiiDegree of membership of, calculatingOr s is greater than or equal to smaxWhether or not it holds, where is the algorithm precision, smaxIs the last round;
step 305, if aboves≥smaxIf the two formulas are not satisfied, the step 302 is skipped to continue the circulation; if one formula is established, the loop ends and the resource thetajAmong the corresponding membership degrees, the membership degree with the maximum value is the resource thetajCorresponding classification;
updating every day according to the load data of the previous dayThe cluster center before the update is represented, and representing the updated cluster center and membership degree, and then updating the algorithm as follows:
when the system is not in initial operation, the membership degree of the resource after the execution of the demand response is corrected, if the system only marks whether the resource reaches the standard, and if the user executes the demand response to several types in the existing classifications, then:
step 401, if the resource is theta'jIf the aging function is not met, the aging function f (t) is:
f(t)=1-e-t
wherein t represents the number of days until the demand response event;
step 402, updating the membership degree of the resource according to the following formula:
2. The demand response resource clustering method according to claim 1, wherein when not initially operating, the membership degree of the resource after the demand response is executed is modified, and if the system expresses the standard reaching condition of the resource in a scoring manner, the method comprises the following steps:
step 501, correcting coefficient f'j(t) the following:
wherein s isjThe score of the corresponding resource is sigma which is the fraction of the resource reaching the standard; f (t) is an aging function; step 502, setting a decision formula delta, and calculating according to the following formula:
step 503, if Δ is greater than or equal to 1, the correction formula of the membership degree is as follows:
if Delta is less than 1, the correction membership is calculated according to the following formula:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710958627.1A CN107656898B (en) | 2017-10-16 | 2017-10-16 | Demand response resource clustering method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710958627.1A CN107656898B (en) | 2017-10-16 | 2017-10-16 | Demand response resource clustering method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107656898A CN107656898A (en) | 2018-02-02 |
CN107656898B true CN107656898B (en) | 2020-11-10 |
Family
ID=61118466
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710958627.1A Active CN107656898B (en) | 2017-10-16 | 2017-10-16 | Demand response resource clustering method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107656898B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109636087B (en) * | 2018-10-24 | 2023-01-24 | 中国电力科学研究院有限公司 | Dynamic clustering method and system for demand response resources |
CN109902949A (en) * | 2019-02-22 | 2019-06-18 | 云南电网有限责任公司 | A kind of demand response resource classification method |
CN116663854B (en) * | 2023-07-24 | 2023-10-17 | 匠人智慧(江苏)科技有限公司 | Resource scheduling management method, system and storage medium based on intelligent park |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103675610A (en) * | 2013-09-29 | 2014-03-26 | 国家电网公司 | Method for extracting characteristic factors in online local discharge detection |
CN104376402A (en) * | 2014-10-31 | 2015-02-25 | 国家电网公司 | Load characteristic classification and synthesis method based on frequency domain indexes |
CN106410781A (en) * | 2015-07-29 | 2017-02-15 | 中国电力科学研究院 | Power consumer demand response potential determination method |
-
2017
- 2017-10-16 CN CN201710958627.1A patent/CN107656898B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103675610A (en) * | 2013-09-29 | 2014-03-26 | 国家电网公司 | Method for extracting characteristic factors in online local discharge detection |
CN104376402A (en) * | 2014-10-31 | 2015-02-25 | 国家电网公司 | Load characteristic classification and synthesis method based on frequency domain indexes |
CN106410781A (en) * | 2015-07-29 | 2017-02-15 | 中国电力科学研究院 | Power consumer demand response potential determination method |
Also Published As
Publication number | Publication date |
---|---|
CN107656898A (en) | 2018-02-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107656898B (en) | Demand response resource clustering method | |
CN109766950B (en) | Industrial user short-term load prediction method based on morphological clustering and LightGBM | |
CN105719045A (en) | Retention risk determiner | |
CN104391879B (en) | The method and device of hierarchical clustering | |
CN108345908A (en) | Sorting technique, sorting device and the storage medium of electric network data | |
CN108268872B (en) | Robust nonnegative matrix factorization method based on incremental learning | |
CN104408479A (en) | Massive image classification method based on deep vector of locally aggregated descriptors (VLAD) | |
CN111325248A (en) | Method and system for reducing pre-loan business risk | |
CN105354228B (en) | Similar diagram searching method and device | |
CN110634288B (en) | Multi-dimensional urban traffic abnormal event identification method based on ternary Gaussian mixture model | |
CN109117715B (en) | SSD (solid State disk) model-based non-component identification method for construction drawing | |
CN107305640A (en) | A kind of method of unbalanced data classification | |
CN108197795A (en) | The account recognition methods of malice group, device, terminal and storage medium | |
CN111104541A (en) | Efficient face picture retrieval method and device | |
CN111340645B (en) | Improved association analysis method for power load | |
CN106919808A (en) | Gene identification system based on change step length least mean square error sef-adapting filter | |
CN113420454A (en) | Environmental capacity obtaining method and device based on atmospheric pollutant standard reaching constraint | |
CN112990583A (en) | Method and equipment for determining mold entering characteristics of data prediction model | |
JPH10143613A (en) | Pattern recognition method | |
CN110704544A (en) | Object processing method, device, equipment and storage medium based on big data | |
CN110084498A (en) | A kind of work dispatching method of server-side, device, computer equipment and storage medium | |
CN107395562A (en) | A kind of financial terminal security protection method and system based on clustering algorithm | |
CN110598995B (en) | Smart client rating method, smart client rating device and computer readable storage medium | |
WO2021022712A1 (en) | Method and device for image recognition model training, and computer device | |
CN109308565B (en) | Crowd performance grade identification method and device, storage medium and computer equipment |
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 |