CN114066068A - Short-term power load prediction method, device, equipment and storage medium - Google Patents

Short-term power load prediction method, device, equipment and storage medium Download PDF

Info

Publication number
CN114066068A
CN114066068A CN202111371538.XA CN202111371538A CN114066068A CN 114066068 A CN114066068 A CN 114066068A CN 202111371538 A CN202111371538 A CN 202111371538A CN 114066068 A CN114066068 A CN 114066068A
Authority
CN
China
Prior art keywords
predicted
load
load prediction
historical
area
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.)
Pending
Application number
CN202111371538.XA
Other languages
Chinese (zh)
Inventor
乡立
段炼
黄锦增
邓祺
王伟超
洪海生
熊俊
林海
陈菁
刘琦
尚明远
林茵茵
魏艳霞
陈永淑
李茜莹
李荣琳
余文铖
唐娴
岳首志
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.)
Guangzhou Power Supply Bureau of Guangdong Power Grid Co Ltd
Original Assignee
Guangzhou Power Supply Bureau of Guangdong Power Grid 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 Guangzhou Power Supply Bureau of Guangdong Power Grid Co Ltd filed Critical Guangzhou Power Supply Bureau of Guangdong Power Grid Co Ltd
Priority to CN202111371538.XA priority Critical patent/CN114066068A/en
Publication of CN114066068A publication Critical patent/CN114066068A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Human Resources & Organizations (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Strategic Management (AREA)
  • Health & Medical Sciences (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Game Theory and Decision Science (AREA)
  • Probability & Statistics with Applications (AREA)
  • Development Economics (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Supply And Distribution Of Alternating Current (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The application discloses a short-term power load prediction method, a device, equipment and a storage medium, wherein the method comprises the following steps: acquiring time to be predicted and a transformer area to be predicted; acquiring a plurality of district load types corresponding to the district to be predicted; acquiring a load prediction model corresponding to each platform area load type; inputting the time to be predicted, the environmental data corresponding to the time to be predicted and the time attribute into each load prediction model to obtain a load prediction value output by each load prediction model; and summing all the load predicted values to obtain a target load predicted value corresponding to the to-be-predicted platform area. The technical problem that the accuracy is low in the existing short-term power load prediction method is solved.

Description

Short-term power load prediction method, device, equipment and storage medium
Technical Field
The present application relates to the field of power load analysis technologies, and in particular, to a method, an apparatus, a device, and a storage medium for short-term power load prediction.
Background
The power load prediction is connected with the power grid energy and the user demand, and the power load can be accurately, timely and effectively predicted to assist the power grid in power scheduling arrangement, so that power accidents are prevented, and large-scale power failure or serious economic loss is avoided. The accurate power grid short-term power load prediction method can realize the fine management of the power grid energy, and is an important guarantee for the stable electricity utilization and the stable economic development of residents.
The short-term power load prediction is to predict the load state of several hours or several days in the future by means of the historical power load fluctuation rule and the influence of external environment data. Although the conventional method can obtain a certain effect on short-term power load prediction, the conventional short-term power load prediction method has low accuracy in practical application.
Disclosure of Invention
In view of this, the present application provides a short-term power load prediction method, apparatus, device and storage medium, which solve the technical problem of low accuracy of the existing short-term power load prediction method.
The application provides a short-term power load prediction method in a first aspect, which includes:
acquiring time to be predicted and a transformer area to be predicted;
acquiring a plurality of district load types corresponding to the district to be predicted;
acquiring a load prediction model corresponding to each platform area load type;
inputting the time to be predicted, the environmental data corresponding to the time to be predicted and the time attribute into each load prediction model to obtain a load prediction value output by each load prediction model;
and summing all the load predicted values to obtain a target load predicted value corresponding to the to-be-predicted platform area.
Optionally, the obtaining of the load types of the plurality of transformer areas corresponding to the transformer area to be predicted specifically includes:
acquiring the area information of the area to be predicted;
and acquiring a plurality of platform area load types corresponding to the platform area information based on the corresponding relation among the platform area information, the platform area information and the plurality of platform area load types.
Optionally, the configuration process of the plurality of platform load types specifically includes:
step S1, acquiring a plurality of historical power loads of the transformer area to be predicted and corresponding categories of the historical power loads;
step S2, randomly selecting q data from the historical power loads as initial clustering centers;
step S3, calculating a distance between each of the historical electrical loads and each of the cluster centers;
step S4, re-determining q clustering centers based on the distance to obtain a clustering center set;
step S5, circularly executing the steps S3 and S4 until the difference value between the current clustering center set and the previous clustering center set is smaller than a preset value;
and step S6, taking the category corresponding to the current clustering center set as the distribution area load type corresponding to the distribution area to be predicted.
Optionally, the configuring process of the load prediction model includes:
acquiring a plurality of historical power loads of the transformer area to be predicted, and historical time attributes and historical environment data corresponding to the historical power loads;
and training a preset initial model by taking the historical time attribute and the historical environment data as input parameters and the historical power load as target output to obtain the load prediction model.
Optionally, the preset initial model is a gradient progressive regression tree model, and the gradient progressive regression tree model is:
Figure BDA0003362512670000021
in the formula (f)0(x) For the gradient progressive regression tree model, argmin () is the penalty function L (y)iC) value of a constant value c when the minimum value is reached, yiIs the ithAnd inputting parameters, wherein N is the number of the input parameters.
Optionally, the method includes training a preset initial model to obtain the load prediction model by using the historical time attribute and the historical environmental data as input parameters and the historical power load as a target output, and the method further includes:
and preprocessing abnormal values of the historical power loads.
A second aspect of the present application provides a short-term power load prediction apparatus, including:
the device comprises a first acquisition unit, a second acquisition unit and a prediction unit, wherein the first acquisition unit is used for acquiring time to be predicted and a station area to be predicted;
the second obtaining unit is used for obtaining a plurality of district load types corresponding to the district to be predicted;
a third obtaining unit, configured to obtain a load prediction model corresponding to each platform area load type;
the prediction unit is used for inputting the time to be predicted, the environmental data corresponding to the time to be predicted and the time attribute into each load prediction model to obtain a load prediction value output by each load prediction model;
and the summing unit is used for summing all the load predicted values to obtain a target load predicted value corresponding to the to-be-predicted platform area.
Optionally, the configuration process of the plurality of platform load types specifically includes:
step S1, acquiring a plurality of historical power loads of the transformer area to be predicted and corresponding categories of the historical power loads;
step S2, randomly selecting q data from the historical power loads as initial clustering centers;
step S3, calculating a distance between each of the historical electrical loads and each of the cluster centers;
step S4, re-determining q clustering centers based on the distance to obtain a clustering center set;
step S5, circularly executing the steps S3 and S4 until the difference value between the current clustering center set and the previous clustering center set is smaller than a preset value;
and step S6, taking the category corresponding to the current clustering center set as the distribution area load type corresponding to the distribution area to be predicted.
A third aspect of the present application provides a short-term power load prediction apparatus comprising a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to perform any of the short term power load prediction methods of the first aspect according to instructions in the program code.
A fourth aspect of the present application provides a storage medium for storing program code for executing the short-term power load prediction method according to any one of the first aspects.
According to the technical scheme, the method has the following advantages:
the application provides a short-term power load forecasting method, which comprises the steps of firstly obtaining time to be forecasted and a station area to be forecasted, then obtaining a plurality of station area load types corresponding to the station area to be forecasted, simultaneously obtaining load forecasting models corresponding to the station area load types, then inputting environmental data and time attributes corresponding to the time to be forecasted and the time to be forecasted into each load forecasting model to obtain load forecasting values output by each load forecasting model, and finally summing all the load forecasting values to obtain a target load forecasting value corresponding to the station area to be forecasted. According to the method and the device, the load prediction is respectively carried out on each load type of the platform area to be predicted, finally, the load prediction values corresponding to each load type are subjected to summation side, the target load prediction value corresponding to the whole platform area to be predicted is obtained, the accuracy of the prediction result is high, and therefore the technical problem that the accuracy is low in the existing short-term power load prediction method is solved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
FIG. 1 is a flowchart illustrating a method for short term power load prediction according to a first embodiment of the present disclosure;
FIG. 2 is a flowchart illustrating a short term power load prediction method according to a second embodiment of the present disclosure;
FIG. 3 is a schematic diagram of a short term power load forecasting method in an embodiment of the present application;
fig. 4 is a schematic structural diagram of a short-term power load prediction apparatus according to an embodiment of the present disclosure.
Detailed Description
The embodiment of the application provides a short-term power load prediction method, a short-term power load prediction device, equipment and a storage medium, and solves the technical problem that the accuracy is low in the conventional short-term power load prediction method.
In order to make the technical solutions of the present application better understood, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. 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.
A first aspect of embodiments of the present application provides an embodiment of a short-term power load prediction method.
Referring to fig. 1, a flowchart of a short-term power load prediction method according to a first embodiment of the present disclosure is shown.
The short-term power load prediction method in the present embodiment includes:
step 101, obtaining time to be predicted and a station area to be predicted.
It can be understood that, in the power load prediction method in the present application, prediction is performed on a power load at a certain time period or a certain time point, so that in the power load prediction, a corresponding time when the power load prediction needs to be obtained, that is, a time to be predicted. It is understood that the specific time information of the time to be predicted may be set according to the prediction requirement, such as a time period, for example, 10/1/2021 to 10/7/2021, or a specific day, for example, 10/9/2021. This is not particularly limited in this embodiment.
And 102, acquiring a plurality of distribution area load types corresponding to the distribution area to be predicted.
In the embodiment, the to-be-predicted distribution area is classified based on the distribution area load types, so that prediction can be performed on various distribution area load types during prediction, the accuracy of prediction results corresponding to various distribution area load types is improved in a targeted prediction mode, and the load prediction accuracy of the whole to-be-predicted distribution area is improved.
It is understood that the above-mentioned type of platform load may be four types of urban residential platform, commercial platform, industrial platform and agricultural platform, or may be other types, which is not specifically limited in this embodiment.
It should be noted that the above-mentioned numbers are indefinite, that is, one, or a plurality (that is, two or more), and specific numerical values of "a number" are specifically determined according to the type of the cell load actually corresponding to the cell to be predicted.
And 103, acquiring a load prediction model corresponding to the load type of each distribution area.
It can be understood that each platform area load type corresponds to one load prediction model, and specifically, when the load prediction model corresponding to the platform area load type is obtained, the load prediction model is obtained according to a corresponding relationship between the platform area load type and the load prediction model. In this embodiment, a platform load type is preset to correspond to a load prediction model, and based on the above-mentioned one-to-one correspondence relationship, the other corresponding to the platform load type can be determined when any one of the platform load type and the load prediction model is known based on the correspondence relationship, that is, when the platform load type is known, the corresponding load prediction model can be determined, or when the load prediction model is known, the corresponding platform load type is known.
And 104, inputting the time to be predicted, the environmental data corresponding to the time to be predicted and the time attribute into each load prediction model to obtain a load prediction value output by each load prediction model.
In order to further improve the load prediction result, in the present embodiment, when performing load prediction, environmental data, time attributes, and the like are also considered. It can be understood that the environmental data may be weather, temperature, humidity, location, area, and the like, and the time attribute may be a working day, a non-working day, a season, and the like, and may be set by a person skilled in the art as needed, which is not specifically limited and described in this embodiment.
Specifically, the load prediction model is obtained by training based on the historical power load and the time and environment corresponding to the historical power load when being configured, so that the power load can be predicted through the load prediction model after the configuration of the load prediction model is completed.
And 105, summing all the load predicted values to obtain a target load predicted value corresponding to the to-be-predicted distribution room.
Since the load prediction in step 104 is performed for each platform load type, after the load prediction values corresponding to each platform load type are obtained, the values are summed up to obtain the target load prediction value corresponding to the platform to be predicted.
In the embodiment, time to be predicted and a platform area to be predicted are firstly obtained, then a plurality of platform area load types corresponding to the platform area to be predicted are obtained, load prediction models corresponding to the load types of the platform areas are simultaneously obtained, then the time to be predicted, environmental data corresponding to the time to be predicted and time attributes are input into the load prediction models to obtain load prediction values output by the load prediction models, and finally all the load prediction values are summed to obtain a target load prediction value corresponding to the platform area to be predicted. According to the method and the device, the load prediction is respectively carried out on each load type of the platform area to be predicted, finally, the load prediction values corresponding to each load type are subjected to summation side, the target load prediction value corresponding to the whole platform area to be predicted is obtained, the accuracy of the prediction result is high, and therefore the technical problem that the accuracy is low in the existing short-term power load prediction method is solved.
The above is a first embodiment of a short-term power load prediction method provided in the embodiments of the present application, and the following is a second embodiment of a short-term power load prediction method provided in the embodiments of the present application.
Referring to fig. 2, a flowchart of a short-term power load prediction method according to a second embodiment of the present application is shown.
The short-term power load prediction method in the present embodiment includes:
step 201, obtaining the time to be predicted and the station area to be predicted.
It should be noted that, the content of step 201 is similar to that of step 101, and reference may be specifically made to the description of step 101, which is not described herein again.
Step 202, obtaining the station area information of the station area to be predicted.
It can be understood that the station area information in this embodiment is used to distinguish the station areas to be predicted, that is, different station areas to be predicted correspond to different station area information (representing station area identities), for example, the station area information may be a station area name, a station area number, and the like, which is not specifically limited in this embodiment
Step 203, acquiring a plurality of platform area load types corresponding to the platform area information based on the corresponding relationship among the platform area information, the platform area information and the plurality of platform area load types.
In this embodiment, the plurality of cell types corresponding to the cell to be predicted are determined based on the correspondence between the cell information of the cell to be predicted and the plurality of cell load types.
Specifically, the configuration process of the load types of the multiple distribution areas in this embodiment specifically includes:
step S1, acquiring a plurality of historical power loads of the transformer area to be predicted and corresponding categories of the historical power loads;
step S2, randomly selecting q data from a plurality of historical power loads as initial clustering centers;
step S3, calculating the distance between each historical power load and each cluster center;
step S4, re-determining q clustering centers based on the distance to obtain a clustering center set;
step S5, circularly executing the steps S3 and S4 until the difference value between the current clustering center set and the previous clustering center set is smaller than a preset value;
and step S6, taking the category corresponding to the current clustering center set as the distribution area load type corresponding to the distribution area to be predicted.
It is understood that the distance calculated in step S3 is different according to the actual problem of clustering, and different distance solving formulas can be selected when solving the distance between two data. Euclidean distance, Manhattan distance, or Minkowski distance can all be used as a measure of "distance" in the algorithm.
Wherein the formula for calculating the manhattan distance or the minkowski distance is as follows:
a. manhattan distance: d (x, y) ═ x1-y1|+|x2-y2|+...+|xn-yn|
b. Minkowski distance:
Figure BDA0003362512670000081
where, when q is 2 and q is 1, the minkowski distance is equal to the euclidean distance and the manhattan distance, respectively.
And 204, acquiring a load prediction model corresponding to the load type of each distribution area.
It is understood that the configuration process of the load prediction model includes:
acquiring a plurality of historical power loads of a transformer area to be predicted, and historical time attributes and historical environment data corresponding to the historical power loads;
and training a preset initial model by taking the historical time attribute and the historical environment data as input parameters and the historical power load as target output to obtain a load prediction model.
The preset initial model is a gradient progressive regression tree model, and the gradient progressive regression tree model is as follows:
Figure BDA0003362512670000082
in the formula (f)0(x) For the gradient progressive regression tree model, argmin () is the penalty function L (y)iC) value of a constant value c when the minimum value is reached, yiIs the ith input parameter, and N is the number of input parameters.
The method comprises the following steps of training a preset initial model by taking historical time attributes and historical environment data as input parameters and historical power loads as target outputs to obtain a load prediction model, wherein the method also comprises the following steps:
and preprocessing abnormal values of the historical power loads.
To ensure that the acquired data is correct and complete, abnormal values need to be processed first, otherwise unnecessary interference is generated to model training. Outliers are characterized by deviations from most of the operational data. Abnormal point detection can be performed by adopting a 3 sigma criterion, if a measured value (namely, a historical time attribute, historical environmental data or historical power load and the like) meets any one of the following two formulas, the measured value is an abnormal value, and the abnormal value is removed, wherein the formula is as follows:
Figure BDA0003362512670000083
wherein the content of the first and second substances,
Figure BDA0003362512670000084
is the average of the measured value history data, sigma is the standard deviation of the measured value history data, xiAre measured values.
To fully utilize the acquired data, the outliers need to be repaired. Because the power load data has slow time variation, the data at adjacent moments are used for interpolation filling, the load variation has strong periodicity, and the load prediction conditions of different types of days are obviously different, so that the data are filled by adopting the average value of the data at the same time and the same type of day in adjacent days.
It should be noted that the characteristic of the gradient-enhanced regression tree (GBRT) algorithm is that it can still obtain the same high precision as the deeper regression tree when the depth of each regression tree is small, and the maximum depth of each regression tree is usually set to a small value to prevent overfitting.
The main parameters of the gradient progressive regression tree (GBRT) model include the maximum depth, learning rate and iteration number of each regression tree. The learning rate is used for controlling the step size of the optimization problem, if the step size is too large, the optimization process may diverge, and if the step size is too small, too many iterations are required to be executed, and the calculation time is increased. According to experience, the learning rate is usually in the range of 0.025-0.3. The iteration number required by model training is related to the difficulty of the prediction problem and the setting of the two parameters, the iteration number is optimized by carrying out cross validation on the existing data, and in the iteration process, when the prediction precision of a validation set is not improved in continuous Q-turn iteration, the iteration is stopped, so that the current iteration number is obtained and is used for final GBRT model training.
And 205, inputting the time to be predicted, the environmental data corresponding to the time to be predicted and the time attribute into each load prediction model to obtain a load prediction value output by each load prediction model.
It is understood that the description of step 205 is the same as that of step 104 in the first embodiment, and reference may be specifically made to the description of step 104, which is not described herein again.
And step 206, summing all the load predicted values to obtain a target load predicted value corresponding to the platform area to be predicted.
It is understood that the description of step 206 is the same as that of step 105 in the first embodiment, and reference may be specifically made to the description of step 105, which is not described herein again.
In the embodiment, time to be predicted and a platform area to be predicted are firstly obtained, then a plurality of platform area load types corresponding to the platform area to be predicted are obtained, load prediction models corresponding to the load types of the platform areas are simultaneously obtained, then the time to be predicted, environmental data corresponding to the time to be predicted and time attributes are input into the load prediction models to obtain load prediction values output by the load prediction models, and finally all the load prediction values are summed to obtain a target load prediction value corresponding to the platform area to be predicted. According to the method and the device, the load prediction is respectively carried out on each load type of the platform area to be predicted, finally, the load prediction values corresponding to each load type are subjected to summation side, the target load prediction value corresponding to the whole platform area to be predicted is obtained, the accuracy of the prediction result is high, and therefore the technical problem that the accuracy is low in the existing short-term power load prediction method is solved.
For easy understanding, please refer to fig. 3, a specific flow of the short-term power load prediction method in the present embodiment is described as follows:
step 1, collecting data historical power load xiAnd data preprocessing is carried out, the phenomena of data loss and data abnormity which often occur in the acquisition and transmission processes of the data are solved, and the correctness and the integrity of the data are ensured.
Step 2, historical power load x is subjected to comparisoniPerforming K-means cluster analysis to obtain power load types 1-q of the whole area of the transformer area to be predicted, wherein the power load of the q-th type of transformer area comprises the numbers 1-NqAnd (4) each platform area.
Carrying out cluster analysis on the power load by adopting a K-means cluster analysis method and carrying out classification treatment, and specifically comprising the following steps:
(1) from a data set
Figure BDA0003362512670000101
Randomly selecting q data as initial clustering center mujWhere N is the number of samples of the historical power load, q0 ═ μ1,μ2,...,μq}；
(2) For the ith sample point x in the data setiCalculating it and each cluster center mujAnd obtaining a sample xiReference numbers of the categories:
Figure BDA0003362512670000102
(3) recalculating the q cluster centers according to the following formula:
Figure BDA0003362512670000103
wherein the content of the first and second substances,
Figure BDA0003362512670000104
is a new cluster center set, where NjThe number of users included in the j-th class.
(4) And (4) repeating the step (2) and the step (3) until the difference value between the current clustering center set and the previous clustering center set is smaller than a preset value. Thereby obtaining a q-type power load classification result of the transformer area, wherein the q-type transformer area load comprises NqAnd (4) each platform area.
Step 3, constructing a feature set input by a model for each type of power load, wherein the feature set mainly comprises historical environment data, historical time attributes, historical time data and the like and can be determined according to specific data conditions and the actual electricity utilization characteristics of regions;
and 4, training a GBRT model and optimizing parameters thereof, constructing a gradient progressive regression tree (GBRT) model aiming at each type of power load, wherein three main parameters in the model comprise the maximum depth, the learning rate and the iteration number of each regression tree.
Let a certain class of training data set be T { (x)1,y1),(x2,y2),...,(xN,yN) In which xi∈Rd,yie.R, set the penalty function to L (y, f (x)), and set the maximum depth of the regression tree to D.
The specific steps for training the GBRT model are as follows:
(1) the model is initialized and a constant value c that minimizes the loss function is calculated according to the following formula, resulting in a tree with only one root node.
Figure BDA0003362512670000111
(2) The value of the negative gradient of the loss function at the current model is calculated as an approximation of the residual error. For M1, 2,., M, N1, 2,., N, the following formula is calculated:
Figure BDA0003362512670000112
(3) fitting with regression trees
Figure BDA0003362512670000113
To obtain the mth regression tree h (x)i,λm) Wherein λ ismIs a model parameter;
(4) model weight beta found by linear searchmThe value of the loss function is minimized.
Figure BDA0003362512670000114
(5) And (5) finishing iteration to obtain a final model:
f(x)＝fM(x)；
and 5, performing data prediction on the power loads of the 1-q types of transformer areas in the transformer area to be predicted to obtain a predicted value of each type of power load.
And respectively carrying out load prediction on the clustered q-type platform district loads by using a GBRT prediction model: first, load N for each type of platform areaqPredicting the power distribution areas respectively to obtain the total power load prediction value of each type of power distribution area
Figure BDA0003362512670000115
And 6, summing the prediction values of the different types of the transformer areas in the step 5 to obtain the predicted value of the medium-short term load of the whole area from days to tens of days in the future.
And finally, synthesizing the load predicted values of each type of the distribution area to obtain the power load predicted value of the whole area.
Figure BDA0003362512670000116
g represents a station class number.
The short-term power load prediction method in the embodiment improves the sensitivity and adaptability of the prediction model to the power load sudden change event, can improve the prediction accuracy of the short-term power load prediction, can enable parameters to be close to the optimal result in a wide range, has strong practicability, and can perform medium-term and short-term prediction on the power load of a whole area.
A second aspect of embodiments of the present application provides an embodiment of a short-term power load prediction apparatus.
Referring to fig. 4, a schematic structural diagram of a short-term power load prediction apparatus according to an embodiment of the present application is shown.
A short-term power load prediction apparatus in this embodiment includes:
the device comprises a first acquisition unit, a second acquisition unit and a prediction unit, wherein the first acquisition unit is used for acquiring time to be predicted and a station area to be predicted;
the second acquisition unit is used for acquiring a plurality of district load types corresponding to the districts to be predicted;
the third acquisition unit is used for acquiring a load prediction model corresponding to the load type of each platform area;
the prediction unit is used for inputting the time to be predicted, the environmental data corresponding to the time to be predicted and the time attribute into each load prediction model to obtain a load prediction value output by each load prediction model;
and the summing unit is used for summing all the load predicted values to obtain a target load predicted value corresponding to the to-be-predicted platform area.
Further, the configuration process of the load types of the plurality of cells specifically includes:
step S1, acquiring a plurality of historical power loads of the transformer area to be predicted and corresponding categories of the historical power loads;
step S2, randomly selecting q data from a plurality of historical power loads as initial clustering centers;
step S3, calculating the distance between each historical power load and each cluster center;
step S4, re-determining q clustering centers based on the distance to obtain a clustering center set;
step S5, circularly executing the steps S3 and S4 until the difference value between the current clustering center set and the previous clustering center set is smaller than a preset value;
and step S6, taking the category corresponding to the current clustering center set as the distribution area load type corresponding to the distribution area to be predicted.
Further, the second obtaining unit specifically includes:
the first acquisition subunit is used for acquiring the station area information of the station area to be predicted;
and the second obtaining subunit is configured to obtain, based on the correspondence between the platform area information, and the plurality of platform area load types, the plurality of platform area load types corresponding to the platform area information.
Further, the configuration process of the load prediction model comprises the following steps:
acquiring a plurality of historical power loads of a transformer area to be predicted, and historical time attributes and historical environment data corresponding to the historical power loads;
and training a preset initial model by taking the historical time attribute and the historical environment data as input parameters and the historical power load as target output to obtain a load prediction model.
Optionally, the preset initial model is a gradient progressive regression tree model, and the gradient progressive regression tree model is:
Figure BDA0003362512670000131
in the formula (f)0(x) For the gradient progressive regression tree model, argmin () is the penalty function L (y)iC) value of a constant value c when the minimum value is reached, yiIs the ith input parameter, and N is the number of input parameters.
Further, the method includes the following steps that historical time attributes and historical environment data are used as input parameters, historical power loads are used as target outputs, a preset initial model is trained, and a load prediction model is obtained, wherein the method also includes the following steps:
and preprocessing abnormal values of the historical power loads.
In the embodiment, time to be predicted and a platform area to be predicted are firstly obtained, then a plurality of platform area load types corresponding to the platform area to be predicted are obtained, load prediction models corresponding to the load types of the platform areas are simultaneously obtained, then the time to be predicted, environmental data corresponding to the time to be predicted and time attributes are input into the load prediction models to obtain load prediction values output by the load prediction models, and finally all the load prediction values are summed to obtain a target load prediction value corresponding to the platform area to be predicted. According to the method and the device, the load prediction is respectively carried out on each load type of the platform area to be predicted, finally, the load prediction values corresponding to each load type are subjected to summation side, the target load prediction value corresponding to the whole platform area to be predicted is obtained, the accuracy of the prediction result is high, and therefore the technical problem that the accuracy is low in the existing short-term power load prediction method is solved.
A third aspect of embodiments of the present application provides an embodiment of a short-term power load prediction apparatus.
A short term power load prediction device comprising a processor and a memory; the memory is used for storing the program codes and transmitting the program codes to the processor; the processor is configured to perform the short term power load prediction method of the first aspect in accordance with instructions in the program code.
A fourth aspect of embodiments of the present application provides an embodiment of a storage medium.
A storage medium for storing program code for performing the short term power load prediction method of the first aspect.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of a unit is only one logical functional division, and there may be other divisions when the actual implementation is performed, for example, a plurality of units or components may be combined or may be integrated into another grid network to be installed, or some features may be omitted or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
Units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.

Claims (10)

1. A method for short-term power load prediction, comprising:
acquiring time to be predicted and a transformer area to be predicted;
acquiring a plurality of district load types corresponding to the district to be predicted;
acquiring a load prediction model corresponding to each platform area load type;
inputting the time to be predicted, the environmental data corresponding to the time to be predicted and the time attribute into each load prediction model to obtain a load prediction value output by each load prediction model;
and summing all the load predicted values to obtain a target load predicted value corresponding to the to-be-predicted platform area.
2. The short-term power load prediction method according to claim 1, wherein the obtaining of the plurality of district load types corresponding to the to-be-predicted district specifically includes:
acquiring the area information of the area to be predicted;
and acquiring a plurality of platform area load types corresponding to the platform area information based on the corresponding relation among the platform area information, the platform area information and the plurality of platform area load types.
3. The short term power load forecasting method as claimed in claim 1, wherein the configuration process of the plurality of the platform load types specifically comprises:
step S1, acquiring a plurality of historical power loads of the transformer area to be predicted and corresponding categories of the historical power loads;
step S2, randomly selecting q data from the historical power loads as initial clustering centers;
step S3, calculating a distance between each of the historical electrical loads and each of the cluster centers;
step S4, re-determining q clustering centers based on the distance to obtain a clustering center set;
step S5, circularly executing the steps S3 and S4 until the difference value between the current clustering center set and the previous clustering center set is smaller than a preset value;
and step S6, taking the category corresponding to the current clustering center set as the distribution area load type corresponding to the distribution area to be predicted.
4. The short term power load prediction method of claim 1, wherein the configuration process of the load prediction model comprises:
acquiring a plurality of historical power loads of the transformer area to be predicted, and historical time attributes and historical environment data corresponding to the historical power loads;
and training a preset initial model by taking the historical time attribute and the historical environment data as input parameters and the historical power load as target output to obtain the load prediction model.
5. The short term power load prediction method according to claim 4, wherein the predetermined initial model is a gradient regression tree model, and the gradient regression tree model is:
Figure FDA0003362512660000021
in the formula (f)0(x) For the gradient progressive regression tree model, argmin () is the penalty function L (y)iC) value of a constant value c when the minimum value is reached, yiIs the ith input parameter, and N is the number of input parameters.
6. The short-term power load prediction method according to claim 4, wherein a preset initial model is trained to obtain the load prediction model by using the historical time attribute and the historical environmental data as input parameters and the historical power load as a target output, and before the method further comprises:
and preprocessing abnormal values of the historical power loads.
7. A short-term power load prediction apparatus, comprising:
the device comprises a first acquisition unit, a second acquisition unit and a prediction unit, wherein the first acquisition unit is used for acquiring time to be predicted and a station area to be predicted;
the second obtaining unit is used for obtaining a plurality of district load types corresponding to the district to be predicted;
a third obtaining unit, configured to obtain a load prediction model corresponding to each platform area load type;
the prediction unit is used for inputting the time to be predicted, the environmental data corresponding to the time to be predicted and the time attribute into each load prediction model to obtain a load prediction value output by each load prediction model;
and the summing unit is used for summing all the load predicted values to obtain a target load predicted value corresponding to the to-be-predicted platform area.
8. The short term power load prediction device of claim 7, wherein the configuration process of the plurality of platform load types specifically comprises:
step S1, acquiring a plurality of historical power loads of the transformer area to be predicted and corresponding categories of the historical power loads;
step S2, randomly selecting q data from the historical power loads as initial clustering centers;
step S3, calculating a distance between each of the historical electrical loads and each of the cluster centers;
step S4, re-determining q clustering centers based on the distance to obtain a clustering center set;
step S5, circularly executing the steps S3 and S4 until the difference value between the current clustering center set and the previous clustering center set is smaller than a preset value;
and step S6, taking the category corresponding to the current clustering center set as the distribution area load type corresponding to the distribution area to be predicted.
9. A short-term power load prediction apparatus comprising a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to execute the short term power load prediction method of any one of claims 1 to 6 in accordance with instructions in the program code.
10. A storage medium for storing program code for performing the short term power load prediction method of any one of claims 1 to 6.
CN202111371538.XA 2021-11-18 2021-11-18 Short-term power load prediction method, device, equipment and storage medium Pending CN114066068A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111371538.XA CN114066068A (en) 2021-11-18 2021-11-18 Short-term power load prediction method, device, equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111371538.XA CN114066068A (en) 2021-11-18 2021-11-18 Short-term power load prediction method, device, equipment and storage medium

Publications (1)

Publication Number Publication Date
CN114066068A true CN114066068A (en) 2022-02-18

Family

ID=80278153

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111371538.XA Pending CN114066068A (en) 2021-11-18 2021-11-18 Short-term power load prediction method, device, equipment and storage medium

Country Status (1)

Country Link
CN (1) CN114066068A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117350485A (en) * 2023-09-27 2024-01-05 广东电网有限责任公司 Power market control method and system based on data mining model

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117350485A (en) * 2023-09-27 2024-01-05 广东电网有限责任公司 Power market control method and system based on data mining model

Similar Documents

Publication Publication Date Title
CN101782743A (en) Neural network modeling method and system
CN110674993A (en) User load short-term prediction method and device
CN112734128A (en) 7-day power load peak value prediction method based on optimized RBF
CN113344288B (en) Cascade hydropower station group water level prediction method and device and computer readable storage medium
CN112733417A (en) Abnormal load data detection and correction method and system based on model optimization
Galvani et al. Data clustering based probabilistic optimal power flow in power systems
CN111723839A (en) Method for predicting line loss rate of distribution room based on edge calculation
CN115796393A (en) Energy network management optimization method, system and storage medium based on multi-energy interaction
CN112990587A (en) Method, system, equipment and medium for accurately predicting power consumption of transformer area
CN114595861A (en) MSTL (modeling, transformation, simulation and maintenance) and LSTM (least Square TM) model-based medium-and-long-term power load prediction method
CN115456306A (en) Bus load prediction method, system, equipment and storage medium
CN115470987A (en) Short-term photovoltaic power generation prediction method based on improved long-term and short-term memory neural network
CN115545333A (en) Method for predicting load curve of multi-load daily-type power distribution network
CN114066068A (en) Short-term power load prediction method, device, equipment and storage medium
CN114154716A (en) Enterprise energy consumption prediction method and device based on graph neural network
CN113780679B (en) Load prediction method and device based on ubiquitous power Internet of things
CN115907131A (en) Method and system for building electric heating load prediction model in northern area
CN113537607B (en) Power failure prediction method
CN116307139A (en) Wind power ultra-short-term prediction method for optimizing and improving extreme learning machine
CN112836885B (en) Combined load prediction method, combined load prediction device, electronic equipment and storage medium
CN115329899A (en) Clustering equivalent model construction method, system, equipment and storage medium
CN112581311B (en) Method and system for predicting long-term output fluctuation characteristics of aggregated multiple wind power plants
CN109086952A (en) It is a kind of based on genetic algorithm-neural network heat load prediction method
CN114139783A (en) Wind power short-term power prediction method and device based on nonlinear weighted combination
Nahid et al. Home occupancy classification using machine learning techniques along with feature selection

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