CN110909786A - New user load identification method based on characteristic index and decision tree model - Google Patents

New user load identification method based on characteristic index and decision tree model Download PDF

Info

Publication number
CN110909786A
CN110909786A CN201911131870.1A CN201911131870A CN110909786A CN 110909786 A CN110909786 A CN 110909786A CN 201911131870 A CN201911131870 A CN 201911131870A CN 110909786 A CN110909786 A CN 110909786A
Authority
CN
China
Prior art keywords
load
node
decision tree
tree model
new user
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.)
Withdrawn
Application number
CN201911131870.1A
Other languages
Chinese (zh)
Inventor
王成亮
官国飞
杨庆胜
钟巍峰
王健
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu Fangtian Power Technology Co Ltd
Original Assignee
Jiangsu Fangtian Power Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jiangsu Fangtian Power Technology Co Ltd filed Critical Jiangsu Fangtian Power Technology Co Ltd
Priority to CN201911131870.1A priority Critical patent/CN110909786A/en
Publication of CN110909786A publication Critical patent/CN110909786A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24323—Tree-organised classifiers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Economics (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a new-installation user load identification method based on characteristic indexes and a decision tree model, which comprises the following steps: s1, calculating load characteristic indexes according to the installation parameters of the business expansion new user, and using the load characteristic indexes as a basis for identifying the load time sequence characteristics of the new user; s2, collecting historical electricity utilization data of a large number of users, calculating load characteristic indexes and class labels to serve as a training data set of a decision tree; s3, generating a CART decision tree model by utilizing the training data set; s4, pruning the decision tree model, reducing complexity and improving identification precision; and S5, inputting the load characteristic indexes of the business expansion new user into the decision tree model, and matching the power utilization mode of the new user to obtain the time sequence characteristics of the new user load. The method can effectively identify and analyze the load time sequence characteristics of the business expansion new users which are not connected with the power grid and have no historical power consumption data.

Description

New user load identification method based on characteristic index and decision tree model
Technical Field
The invention relates to a new-installation user load identification method based on characteristic indexes and a decision tree model, and belongs to the technical field of load characteristic analysis of power systems.
Background
The determination of the load characteristics of the users has been a very important topic for the grid companies to support various load management functions. In the traditional business expansion and installation business, information provided by a large user applying for new installation and access is only limited to list information and limited load indexes of maximum load electric equipment, so that an electric power enterprise cannot accurately grasp the electric power load characteristics of the large user, and the electric power load resources are not fully utilized. The lack of this information is very likely to pose serious challenges to the efficient operation of the power distribution network. The method is characterized in that a large number of homogeneous user loads are accessed in a certain power supply point in a concentrated manner, so that the load peak-valley difference of the power supply point is increased continuously, and the capacity transmission limit of equipment is easy to be reached at the peak load time; meanwhile, the method also presents a series of problems of low utilization rate of equipment, capacity waste, poor economy and the like in a long time period. Therefore, it is important to identify the load timing characteristics of the new subscriber before the new subscriber in the cell accesses.
With the wide application of the power demand side management technology, identifying the load time sequence characteristics of the users has important significance for economic analysis, stable operation and power planning of the power system. At present, a cluster analysis theory is mostly adopted, a gray relevance matrix is constructed through gray relevance clustering, information entropy segmentation aggregation approximation and other methods are used for identifying and analyzing load time sequence characteristics of users of an electric power system, the users in the electric power system are divided into different power utilization categories based on historical data, corresponding power utilization labels are established, the load types of unknown users are predicted through the classified user corresponding labels, and an approximate load curve of a new user is obtained. However, for the business expansion new user, because the power grid is not connected and the historical electricity consumption data is not available, the time sequence characteristics of the load cannot be mined through the historical electricity consumption data, and the current identification method for the load time sequence characteristics of the business expansion new user is difficult.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a new user load identification method based on characteristic indexes and a decision tree model so as to solve the problem that the load time sequence characteristics of a new user are difficult to identify in the prior art.
In order to solve the technical problems, the technical scheme of the invention is as follows:
a new user load identification method based on characteristic indexes and a decision tree model comprises the following steps:
calculating according to the installation parameters of the business expansion new user to obtain a load characteristic index;
and inputting the load characteristic index into a high-precision decision tree model to obtain the load time sequence characteristics of the new user.
Further, the reporting parameters comprise estimated daily load power peak value, daily load power base value, estimated whole-day power consumption, estimated peak power consumption, estimated valley power consumption and estimated average power consumption.
Further, the method for calculating the load characteristic index includes:
Figure BDA0002278812250000021
wherein, PmaxTo estimate the daily load power peak, PminIs a daily load power base value, Q is estimated total daily power consumption, QpeakFor estimating peak power consumption, QvalElectricity consumption in the valley period, QshFor flat-term power consumption, PavTo estimate the daily load average power, Pav.peakIs the peak load average power, Pav.shFor average power of the load during the flat period, Pav.valThe load average power during the valley period.
Further, the method for establishing the decision tree model comprises the following steps:
calculating to obtain a historical load characteristic index and a category label according to the electricity utilization data of the historical user;
and taking the historical load characteristic index and the class label as a training data set to train and generate a decision tree model.
And pruning the decision tree model to obtain a high-precision decision tree model.
Further, the training generation process is as follows:
putting the training data set into a root node of the decision tree, and calculating a kini coefficient of the impurity degree of the root node;
splitting the root node according to the load characteristic index to obtain a left child node and a right child node;
calculating to obtain the impurity reduction amount after splitting according to the kini coefficients of the left child node and the right child node;
selecting a load characteristic index corresponding to the maximum impurity reduction amount as a splitting attribute, and splitting the root node again;
continuously splitting the newly split left and right child nodes according to the splitting attribute;
and repeating the process until all the leaf nodes have the kini coefficients smaller than the threshold value, and obtaining the decision tree.
Further, the calculation method of the kini coefficient of the root node is as follows;
Figure BDA0002278812250000031
Figure BDA0002278812250000032
in the formula, GINI(n1)Is a root node n1P (X) is the degree of impurityi|n1) Indicates the load pattern XiThe proportion of the training data set Z, ZN is the number of samples of the training data set, L is the number of user mode categories in the training data set, ZiIs a load mode XiThe number of samples.
Further, the method for calculating the reduction amount of the impurities is as follows;
Φ(n1)＝GINI(n1)-p1-2GINI(n2)-p1-3GINI(n3) (4)
in the formula, phi (n)1) For impurity reduction, p1-2Is a root node n1Is divided into sub-nodes n2The probability of (2) being higher than (b),p1-3is a root node n1Is divided into sub-nodes n3Probability of being in, GINI (n)2) Is the second child node's Kearny coefficient, GINI (n)3) Is the third child node's kini coefficient;
further, the pruning process is as follows:
calculating error gains of non-leaf nodes in the decision tree;
and acquiring the non-leaf node with the minimum error gain and cutting off the subtree of the node.
Further, the error gain is calculated as follows:
Figure BDA0002278812250000033
where α is the error gain of the non-leaf node, R (T)t) Is the error cost before node t pruning, | NTtI is the number of leaf nodes in the subtree, R (T) is the subtree TtAfter pruned, it becomes the error cost of node t of the leaf node.
Further, the error cost is calculated as follows:
R(t)＝r(t)*p(t) (6)
wherein, R (t) error cost, r (t) is error rate of the node t, which is the proportion of misclassified data in the node t to the total data in the node t; p (t) is the proportion of data on node t to all data.
Compared with the prior art, the invention has the following beneficial effects:
the method adopts the decision tree model to identify the time sequence characteristics of the user load, calculates the load characteristic index of the new user through the installation parameters of the new user, inputs the load characteristic index of the new user into the decision tree model, identifies the power consumption mode of the new user to obtain the time sequence characteristics of the new user load, and solves the problem that the prior art cannot identify the time sequence characteristics of the new user load for business expansion.
Drawings
FIG. 1 is a flow chart of a method of an embodiment of the present invention;
FIG. 2 is a CART decision tree power pattern recognition rule diagram.
Detailed Description
For a fuller understanding and appreciation of the invention, the same will be further described in connection with the accompanying drawings and detailed description:
the invention provides a new-installed user load identification method based on characteristic indexes and a decision tree model, the flow of the method is shown in figure 1, and as can be seen from figure 1, the method comprises the following steps:
step S1, load characteristic indexes are calculated according to the installation parameters of the business expansion new user and are used as the basis for identifying the load time sequence characteristics of the new user;
step S2, collecting historical electricity consumption data of a large number of users, calculating load characteristic indexes and class labels to serve as a training data set of a decision tree;
step S3, generating a CART decision tree model by utilizing a training data set;
step S4, performing tree pruning on the decision tree model to obtain a high-precision decision tree model, reducing complexity and improving identification precision;
and step S5, inputting the load characteristic index of the business expansion new user into the decision tree model, and matching the power consumption mode of the new user to obtain the time sequence characteristics of the new user load.
Further, in step S1, specifically, the method includes: and calculating a load characteristic index according to the installation parameters of the business expansion new user as a basis for identifying the load time sequence characteristics of the new user.
In the application service, the application parameters of the new user include: (1) pre-estimated daily load power peak value PmaxDaily load power basic value Pmin(ii) a (2) Estimating the power consumption Q of the whole day; (3) estimation of peak power consumption QpeakElectricity consumption Q in valley periodvalElectric quantity Q for flat periodsh. According to the installation parameters of the new user, the following load characteristic indexes can be calculated:
Figure BDA0002278812250000051
wherein, PmaxTo estimate the daily load power peak, PminIs a daily load power base value, Q is estimated total daily power consumption, QpeakFor estimating peak power consumption, QvalElectricity consumption in the valley period, QshFor flat-term power consumption, PavTo estimate the daily load average power, Pav.peakIs the peak load average power, Pav.shFor average power of the load during the flat period, Pav.valThe load average power during the valley period.
Further, in step S2, specifically, the method includes: and collecting historical electricity utilization data of massive users, and calculating load characteristic indexes and class labels to serve as a training data set of the decision tree.
Typical daily load curves of a plurality of users are collected to serve as a sample data set, load characteristic indexes of the samples are calculated, a complete user load characteristic library is constructed according to a power system, and the power utilization mode types of the samples are matched. And forming a training data set of the decision tree by taking the load characteristic index of the sample data set as a characteristic value and taking the class of the power utilization mode as a class label value.
Further, in step S3, specifically, the method includes: generating a CART decision tree model using the training data set.
Putting a training data set Z into a root node n of a CART decision tree1Its complexity coefficient is calculated. In the training data set, if it belongs to the load pattern XiThe number of samples of (1) is ziAnd then the current root node impurity degree Keyney coefficient is as follows:
Figure BDA0002278812250000061
Figure BDA0002278812250000062
in the formula, p (X)i|n1) Indicates the load pattern XiZN is the number of samples in the training data set, and L is the number of user mode categories in the training data set.
According to a certain load characteristic index, the root node n is aligned1To carry outSplitting to generate two sub-nodes n2And n3Their Gini coefficient is GINI (n)2) And GINI (n)3) The reduction phi (n) of the impurity degree after the splitting is calculated1) Comprises the following steps:
Φ(n1)＝GINI(n1)-p1-2GINI(n2)-p1-3GINI(n3) (4)
in the formula, p1-2Is a root node n1Is divided into sub-nodes n2Probability of (1), p1-3Is a root node n1Is divided into sub-nodes n3Is determined. Aiming at different load characteristic indexes, the calculated impurity reduction amounts are different, the load characteristic index with the largest reduction amount is selected as a splitting attribute, and the root node is split into a left child node and a right child node. And then, taking the two newly split sub-nodes as root nodes of the left and right subtrees, splitting again by using the method, so that the decision tree continuously grows until the kini coefficients of all leaf nodes are smaller than a threshold value, and obtaining the maximum tree T0 of the CART decision tree.
Further, in step S4, specifically, the method includes: and (4) carrying out tree pruning on the decision tree model, reducing complexity and improving identification precision.
After the maximum tree T0 is generated, a set of validation data sets with known eigenvalues and class label values are substituted into the decision tree model, each non-leaf node in the decision tree divides the data set into two according to a certain eigenvalue, and finally, class identification results for the data are obtained at the leaf nodes, and the error gain α of the non-leaf nodes in the CART decision tree is calculated.
Figure BDA0002278812250000071
In the formula, R (T)t) Is the error cost before node T pruning, which is equal to sub-tree TtThe sum of the error costs of all the leaf nodes;
Figure BDA0002278812250000072
the number of leaf nodes in the subtree;r (T) is a subtree TtAfter pruned, it becomes the error cost of node t of the leaf node. The error cost R (t) can be calculated as follows:
R(t)＝r(t)*p(t) (6)
wherein, r (t) is the error rate of the node t, which is the proportion of the misclassified data in the node t to the total data in the node t, and p (t) is the proportion of the data on the node t to all the data, after α values of all non-leaf nodes are obtained through calculation, the non-leaf node with the minimum α value is found, a sub-tree of the node is cut off, and the CART decision tree with the minimum error rate can be obtained through pruning.
Further, in step S5, specifically, the method includes: and inputting the load characteristic indexes of the business expansion new user into the decision tree model, and matching the power utilization mode of the new user to obtain the time sequence characteristics of the load of the new user.
And inputting the load characteristic indexes of the business expansion new user into the decision tree model, comparing the corresponding load characteristic indexes from the root node of the decision tree according to the judgment rule of each node, and selecting corresponding child nodes according to the comparison result. And repeating the steps till the leaf node, and taking the power utilization pattern in the leaf node as a matching result of the power utilization pattern of the new user, thereby obtaining the time sequence characteristics of the load of the new user.
In the embodiment of the invention, the load time sequence characteristics of 5 new reporting users are identified, and the constructed pruned CART decision tree model is shown in FIG. 2. X1 to x6 in fig. 2 represent 6 load characteristic indexes of the user. According to the rule in fig. 2, the power consumption mode categories of the new installation users can be quickly identified by comparing the values of the 6 load characteristic indexes of the new installation users respectively. Through CART decision tree identification, 5 new reporting users correspond to the power utilization mode 1, the power utilization mode 2, the power utilization mode 3 and the power utilization mode 5 respectively.
After the power consumption mode of the new reporting user is known, a complete user load characteristic library is constructed in the power system, and the load time sequence characteristics of the new user can be obtained and represented by typical daily load data of the new user. Typical daily load data in the range of 10:00 to 12:00 obtained after identification of 5 new package users in the embodiment of the invention is shown in table 1:
TABLE 1 New Provisioning of typical daily load data for users
Figure BDA0002278812250000081
The foregoing detailed description has described the present application, and the present application uses specific examples to explain the principles and embodiments of the present application, and the description of the embodiments is only used to help understand the method and core ideas of the present application, and all changes can be made in the specific embodiments and application scope, so in summary, the present application should not be construed as limiting the present application.

Claims (10)

1. A new user load identification method based on characteristic indexes and a decision tree model is characterized by comprising the following steps:
calculating according to the installation parameters of the business expansion new user to obtain a load characteristic index;
and inputting the load characteristic index into a high-precision decision tree model to obtain the load time sequence characteristics of the new user.
2. The method of claim 1, wherein the reporting parameters include an estimated daily load power peak value, a daily load power base value, an estimated total daily power consumption, an estimated peak power consumption, a valley power consumption, and a flat power consumption.
3. The method according to claim 1, wherein the calculating method of the load characteristic index comprises:
Figure FDA0002278812240000011
wherein, PmaxTo estimate the daily load power peak, PminIs a daily load power base value, Q is estimated total daily power consumption, QpeakFor estimating peak power consumption, QvalElectricity consumption in the valley period, QshFor flat-term power consumption, PavTo estimate the daily load average power, Pav.peakIs the peak load average power, Pav.shFor average power of the load during the flat period, Pav.valThe load average power during the valley period.
4. The method according to claim 1, wherein the method for identifying the load of the newly-installed user based on the characteristic index and the decision tree model comprises:
calculating to obtain a historical load characteristic index and a category label according to the electricity utilization data of the historical user;
and taking the historical load characteristic index and the class label as a training data set to train and generate a decision tree model.
And pruning the decision tree model to obtain a high-precision decision tree model.
5. The method according to claim 4, wherein the training generation process comprises:
putting the training data set into a root node of the decision tree, and calculating a kini coefficient of the impurity degree of the root node;
splitting the root node according to the load characteristic index to obtain a left child node and a right child node;
calculating to obtain the impurity reduction amount after splitting according to the kini coefficients of the left child node and the right child node;
selecting a load characteristic index corresponding to the maximum impurity reduction amount as a splitting attribute, and splitting the root node again;
continuously splitting the newly split left and right child nodes according to the splitting attribute;
and repeating the process until all the leaf nodes have the kini coefficients smaller than the threshold value, and obtaining the decision tree.
6. The method according to claim 5, wherein the root node has a kini coefficient calculated as follows;
Figure FDA0002278812240000021
Figure FDA0002278812240000022
in the formula, GINI(n1)Is a root node n1P (X) is the degree of impurityi|n1) Indicates the load pattern XiThe proportion of the training data set Z, ZN is the number of samples of the training data set, L is the number of user mode categories in the training data set, ZiIs a load mode XiThe number of samples.
7. The method according to claim 5, wherein the impurity reduction is calculated as follows;
Φ(n1)＝GINI(n1)-p1-2GINI(n2)-p1-3GINI(n3) (4)
in the formula, phi (n)1) For impurity reduction, p1-2Is a root node n1Is divided into sub-nodes n2Probability of (1), p1-3Is a root node n1Is divided into sub-nodes n3Probability of being in, GINI (n)2) Is the second child node's Kearny coefficient, GINI (n)3) Is the kini coefficient of the third child node.
8. The method according to claim 1, wherein the pruning process comprises:
calculating error gains of non-leaf nodes in the decision tree;
and acquiring the non-leaf node with the minimum error gain and cutting off the subtree of the node.
9. The method according to claim 8, wherein the error gain is calculated as follows:
Figure FDA0002278812240000031
where α is the error gain of the non-leaf node, R (T)t) Is the error cost before node t prunes,
Figure FDA0002278812240000032
is the number of leaf nodes in the sub-tree, R (T) is the sub-tree TtAfter pruned, it becomes the error cost of node t of the leaf node.
10. The method of claim 8, wherein the error cost is calculated as follows:
R(t)＝r(t)*p(t) (6)
wherein, R (t) error cost, r (t) is error rate of the node t, which is the proportion of misclassified data in the node t to the total data in the node t; p (t) is the proportion of data on node t to all data.
CN201911131870.1A 2019-11-19 2019-11-19 New user load identification method based on characteristic index and decision tree model Withdrawn CN110909786A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911131870.1A CN110909786A (en) 2019-11-19 2019-11-19 New user load identification method based on characteristic index and decision tree model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911131870.1A CN110909786A (en) 2019-11-19 2019-11-19 New user load identification method based on characteristic index and decision tree model

Publications (1)

Publication Number Publication Date
CN110909786A true CN110909786A (en) 2020-03-24

Family

ID=69817939

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911131870.1A Withdrawn CN110909786A (en) 2019-11-19 2019-11-19 New user load identification method based on characteristic index and decision tree model

Country Status (1)

Country Link
CN (1) CN110909786A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111915056A (en) * 2020-06-16 2020-11-10 广东电网有限责任公司 User practical load prediction system and prediction method based on big data analysis
CN112487033A (en) * 2020-11-30 2021-03-12 国网山东省电力公司电力科学研究院 Service visualization method and system for data flow and network topology construction
CN113516297A (en) * 2021-05-26 2021-10-19 平安国际智慧城市科技股份有限公司 Prediction method and device based on decision tree model and computer equipment

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111915056A (en) * 2020-06-16 2020-11-10 广东电网有限责任公司 User practical load prediction system and prediction method based on big data analysis
CN112487033A (en) * 2020-11-30 2021-03-12 国网山东省电力公司电力科学研究院 Service visualization method and system for data flow and network topology construction
CN113516297A (en) * 2021-05-26 2021-10-19 平安国际智慧城市科技股份有限公司 Prediction method and device based on decision tree model and computer equipment
CN113516297B (en) * 2021-05-26 2024-03-19 平安国际智慧城市科技股份有限公司 Prediction method and device based on decision tree model and computer equipment

Similar Documents

Publication Publication Date Title
CN107800140B (en) Large user power supply access decision method considering load characteristics
CN106446967A (en) Novel power system load curve clustering method
CN110909786A (en) New user load identification method based on characteristic index and decision tree model
CN107528350B (en) A kind of wind power output typical scene generation method adapting to long -- term generation expansion planning
CN110825723B (en) Resident user classification method based on electricity load analysis
CN106485089B (en) The interval parameter acquisition methods of harmonic wave user's typical condition
CN111552813A (en) Power knowledge graph construction method based on power grid full-service data
CN110380444B (en) Capacity planning method for distributed wind power orderly access to power grid under multiple scenes based on variable structure Copula
CN108960586B (en) Non-invasive load identification method adaptive to scene change
CN115618249A (en) Low-voltage power distribution station area phase identification method based on LargeVis dimension reduction and DBSCAN clustering
CN117559443A (en) Ordered power utilization control method for large industrial user cluster under peak load
CN112865089A (en) Improved large-scale scene analysis method for active power distribution network
CN116821832A (en) Abnormal data identification and correction method for high-voltage industrial and commercial user power load
CN113595071A (en) Transformer area user identification and voltage influence evaluation method
CN112651576A (en) Long-term wind power prediction method and device
CN116796403A (en) Building energy saving method based on comprehensive energy consumption prediction of commercial building
CN116470491A (en) Photovoltaic power probability prediction method and system based on copula function
CN115186882A (en) Clustering-based controllable load spatial density prediction method
CN113191656B (en) Low-voltage distribution network equipment load and topology linkage method based on data correlation analysis
CN114528284A (en) Bottom layer data cleaning method and device, mobile terminal and storage medium
CN112766590B (en) Method and system for extracting typical residential power consumption pattern
CN109858667A (en) It is a kind of based on thunder and lightning weather to the short term clustering method of loading effects
CN111651448A (en) Low-voltage topology identification method based on noise reduction differential evolution
CN114781685B (en) Large user electricity load prediction method and system based on big data mining technology
CN114238045A (en) System and method for judging and automatically repairing integrity of multi-source measurement data of power grid

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
WW01 Invention patent application withdrawn after publication

Application publication date: 20200324

WW01 Invention patent application withdrawn after publication