CN111191811A - Cluster load prediction method and device and storage medium - Google Patents

Cluster load prediction method and device and storage medium Download PDF

Info

Publication number
CN111191811A
CN111191811A CN201811350529.0A CN201811350529A CN111191811A CN 111191811 A CN111191811 A CN 111191811A CN 201811350529 A CN201811350529 A CN 201811350529A CN 111191811 A CN111191811 A CN 111191811A
Authority
CN
China
Prior art keywords
cluster
load
individual
predicted
data
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
CN201811350529.0A
Other languages
Chinese (zh)
Inventor
黎昱
汪建
汤奕
崔晗
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.)
ZTE Corp
Original Assignee
ZTE Corp
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 ZTE Corp filed Critical ZTE Corp
Priority to CN201811350529.0A priority Critical patent/CN111191811A/en
Priority to PCT/CN2019/118309 priority patent/WO2020098728A1/en
Publication of CN111191811A publication Critical patent/CN111191811A/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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • 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)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Economics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biophysics (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • Operations Research (AREA)
  • Primary Health Care (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Biomedical Technology (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Supply And Distribution Of Alternating Current (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a cluster load prediction method, a device and a storage medium, wherein the method comprises the following steps: acquiring historical electricity consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data in the same period; learning and fitting individual historical electricity utilization data and external data in the same period through a machine learning algorithm to form an individual electricity utilization load prediction model; the method comprises the steps of obtaining external data of a cluster to be predicted, predicting the total load of the cluster to be predicted through an individual power load prediction model based on the external data of the cluster to be predicted, and predicting the total load of the cluster to be predicted by combining the individual power load prediction model and an error correction model. The invention improves the refinement degree and the prediction effect of the cluster load prediction.

Description

Cluster load prediction method and device and storage medium
Technical Field
The invention relates to the technical field of electric power, in particular to a cluster load prediction method and device based on distributed power utilization data and external information and a storage medium.
Background
The load prediction of the power system is very important content for operation scheduling and power generation planning of the power system, is an important aspect for ensuring safe and economic operation of the power system and realizing scientific management and scheduling of a power grid, is a component of an energy management system, and is necessary content for carrying out commercial operation of the power grid in the future. The traditional research is to analyze and predict from the total load and external data, but has the problem that details are roughly ignored in prediction, and the refinement degree of cluster load prediction is influenced.
Disclosure of Invention
The invention mainly aims to provide a cluster load prediction method, a cluster load prediction device and a storage medium, aiming at improving the refinement degree and the prediction effect of cluster load prediction.
In order to achieve the above object, the present invention provides a method for predicting a cluster load, including:
acquiring historical electricity consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data in the same period;
learning and fitting the historical electricity utilization data of the individual and the external data of the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual;
acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value.
Optionally, the step of obtaining external data of the cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and through the power load prediction model of the individual, and outputting a cluster load prediction value further includes:
establishing an error correction model between the individual power load and the cluster total load;
the steps of obtaining external data of the cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value comprise:
and predicting the total load of the cluster to be predicted by combining the individual power load prediction model and the error correction model, and outputting a cluster load predicted value.
Optionally, the step of creating an error correction model between the individual power load and the cluster total load includes:
acquiring historical total load of a cluster to which the individual belongs in the same period as the historical electricity consumption data of each individual, and taking the historical total load as the total load of the sample cluster corresponding to the cluster to be predicted;
predicting the power load of the individuals in the sample cluster corresponding to the cluster to be predicted based on the power load prediction model of the individuals to obtain the power load prediction value of each individual;
accumulating the predicted value of the power load of each individual to obtain the predicted accumulated value of the load of the individual;
and analyzing the data association relation between the load prediction accumulated value of the individual and the total load of the sample cluster to form an error correction model between the power load of the individual and the total load of the cluster.
Optionally, after the step of obtaining the historical electricity consumption data of each individual in the sample cluster corresponding to the cluster to be predicted and the external data of the same period, the method further includes:
preprocessing the historical electricity utilization data of the individual and the external data in the same period;
the step of forming an individual power load prediction model by fitting the historical power consumption data of the individual to the external data of the same period through machine learning algorithm learning comprises the following steps:
and learning and fitting the preprocessed individual historical electricity utilization data and the external data in the same period through a machine learning algorithm to form an individual electricity utilization load prediction model.
Optionally, the external data includes: one or more of meteorological data, economic data, social information; the individual power load prediction model is a neural network model or a linear regression model; the error correction model is a linear regression model or an extreme learning machine model.
Optionally, the method further comprises:
outputting the cluster load predicted value to an electric power service platform to provide data support for the electric power service platform; or detecting abnormal electricity utilization behaviors based on the cluster load predicted value.
In addition, an embodiment of the present invention further provides a cluster load prediction apparatus, including:
the data acquisition module is used for acquiring historical electricity consumption data of each individual in a sample cluster corresponding to the cluster to be predicted and external data in the same period;
the load prediction model creating module is used for learning and fitting the historical electricity utilization data of the individual and the external data in the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual;
and the prediction module is used for acquiring external data of the cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and outputting a cluster load prediction value through the power load prediction model of the individual.
Optionally, the apparatus further comprises:
the error correction model creating module is used for creating an error correction model between the individual power load and the cluster total load;
the prediction module is further configured to acquire external data of the cluster to be predicted, predict the total load of the cluster to be predicted based on the external data of the cluster to be predicted and combine the power load prediction model and the error correction model of the individual, and output a cluster load prediction value.
The embodiment of the present invention further provides a device for predicting cluster load, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the steps of the cluster load prediction method as described above.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the cluster load prediction method described above are implemented.
According to the cluster load prediction method, the cluster load prediction device and the storage medium, historical power consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data of the same period are obtained; learning and fitting individual historical electricity utilization data and external data in the same period through a machine learning algorithm to form an individual electricity utilization load prediction model; acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted through an individual power load prediction model based on the external data of the cluster to be predicted, outputting a cluster load prediction value, realizing quick and accurate judgment of abnormal users, and providing data support for power services such as power scheduling and power generation planning; in addition, the total load of the cluster to be predicted can be predicted by combining an individual power load prediction model and an error correction model. Compared with the prior art, the large task of cluster load prediction is decomposed into load prediction subtasks for a large number of individuals, refined measurement data in the power system are fully utilized, so that the microscopic research and analysis on individual data provide help for macroscopic load prediction, more measurement data are considered in the load prediction, the refinement degree of the cluster load prediction is improved by utilizing refined influence factors, data errors between the individuals and the cluster loads are mined through a machine learning algorithm, the load change trend is comprehensively considered by combining the individual power consumption data and external information such as weather and holidays, and the prediction effect of the cluster loads is improved; in addition, the load forecasting speed is accelerated by utilizing the machine learning algorithm, and the short-term load forecasting can be rapidly updated.
Drawings
FIG. 1 is a schematic flow chart of a cluster load prediction method according to a first embodiment of the present invention;
FIG. 2 is a flowchart illustrating a cluster load prediction method according to a second embodiment of the present invention;
FIG. 3 is a flowchart illustrating a cluster load prediction method according to a third embodiment of the present invention;
FIG. 4 is a functional block diagram of cluster load prediction according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a cluster load prediction process according to an embodiment of the present invention;
FIG. 6 is a schematic diagram illustrating the comparison between the load prediction and the actual measurement of the residential users in the cluster load prediction method according to the embodiment of the present invention;
FIG. 7 is a schematic diagram illustrating the comparison of the prediction accuracy in the test data with other methods in the cluster load prediction method according to the embodiment of the present invention;
FIG. 8 is a schematic diagram of an artificial neural network according to example 1 of the present invention;
FIG. 9 is a graph illustrating test results from a test using the artificial neural network of FIG. 8;
FIG. 10 is a schematic diagram illustrating comparison of the accumulated predicted results with the actual power consumption of the cluster load for error correction of the cluster user load in example 1 of the present invention;
FIG. 11 is a graph showing the results of error correction of the data of FIG. 10;
FIG. 12 is a graph showing the regression prediction effect of data in the training of the individual load prediction algorithm in example 2 of the present invention;
FIG. 13 is a schematic diagram of a cluster user load error correction test result in example 2 of the present invention;
FIG. 14 is a graph showing the results of individual load prediction in example 3 of the present invention;
FIG. 15 is a schematic diagram of a cluster user load error correction test result in example 3 of the present invention;
FIG. 16 is a schematic diagram of a cluster user load error correction test result in example 4 of the present invention;
fig. 17 is a schematic structural diagram of an apparatus according to an embodiment of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
In order to make the technical solution of the present invention clearer and clearer, the following detailed description is made with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The main solution of the embodiment of the invention is as follows: historical electricity utilization data of each individual in a sample cluster corresponding to a cluster to be predicted and external data in the same period are obtained; learning and fitting individual historical electricity utilization data and external data in the same period through a machine learning algorithm to form an individual electricity utilization load prediction model; the method is characterized by predicting the total load of the cluster to be predicted by combining an individual power load prediction model and an error correction model, fully utilizing refined measurement data in the power system, mining data errors between the individual and the cluster load through a machine learning algorithm, combining the individual power data and external information such as weather and holidays, comprehensively considering the load change trend, improving the refinement degree and the prediction effect of the cluster load prediction, realizing the rapid and accurate judgment of abnormal users, and providing data support for power services such as power scheduling and power generation planning.
In the prior art, analysis and prediction are performed from the total load and external data, but the problem that details are roughly ignored in prediction exists, and the refinement degree of cluster load prediction is influenced.
The invention provides a solution which can improve the refinement degree and the prediction effect of the cluster load prediction. The method mainly utilizes the user electricity utilization information collected by the intelligent electric meter and external information such as weather, holidays and the like, and comprehensively considers the load change trend. The method relates to data analysis related technologies, including relevance analysis, numerical relation fitting, error analysis and the like. The common algorithms comprise a neural network, a support vector machine, an extreme learning machine, a decision tree and the like, the large task of cluster load prediction is decomposed into load prediction subtasks for a large number of individuals, refined measurement data in an electric power system are fully utilized, then assistance is provided for macroscopic load prediction through microscopic research and analysis of individual data, more measurement data are considered in load prediction, the prediction refinement degree and the prediction effect of the cluster load are improved, and the abnormal users are quickly and accurately judged.
Specifically, as shown in fig. 1, a first embodiment of the present invention provides a cluster load prediction method, including:
step S101, acquiring historical electricity consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data of the same period;
the scheme of the embodiment relates to forecasting of cluster loads in an electric power system, micro load forecasting is conducted on individuals through a machine learning algorithm based on distributed power utilization data and external data, a cluster load forecasting result is obtained through analysis of the relation between the individual load forecasting result and the overall load data of a cluster, and quick and accurate judgment on abnormal users is achieved.
Specifically, firstly, historical electricity consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data of the same period are obtained.
The sample cluster corresponding to the cluster to be predicted can be a cluster of all individuals in a certain region, the power consumption data of each individual in the sample cluster corresponding to the cluster to be predicted is distributed power consumption data, historical power consumption data of a certain time period can be counted according to prediction requirements, and the historical power consumption data can be used as a basis for predicting the power consumption load of the cluster to be predicted in a certain time period in the future.
The distributed power consumption data generally refers to individual power consumption data in the whole power consumption, and includes residential users, commercial districts, provincial administrative units and nationwide. Therefore, the application range of the proposed prediction method covers such dependent electricity consumption individuals and the whole, and the whole (cluster) electricity load is predicted by the individual electricity load.
Therefore, the historical total load of the cluster to which the individual belongs in the same period as the historical electricity consumption data of each individual is acquired as the total load of the sample cluster corresponding to the cluster to be predicted.
Then, historical electricity utilization data of each individual and external data of the same period are obtained from the sample cluster overall load.
Such as: the sample cluster is 873 users in a certain area from 6 months to 7 months, the total load of the sample cluster is the total power consumption load of 873 users in the certain area from 6 months to 7 months, and the historical power consumption data of each individual is the historical power consumption data of each user in 873 users in the certain area from 6 months to 7 months.
The cluster to be predicted is 873 users in a certain area in 8 months, and the total load of the cluster to be predicted is the total power load of 873 users in a certain area in 8 months based on actual data of a sample cluster (873 users in a certain area in 6 months to 7 months).
The rule for selecting the sample cluster corresponding to the cluster to be predicted may be set according to an actual situation, for example, a cluster separated from the cluster to be predicted by a preset time period (for example, 1 month, 10 months, and the like) may be selected as the sample cluster, for example, the cluster to be predicted is a cluster user in 2013 to 2018, and the sample cluster is a cluster user in 2003 to 2012.
In addition, considering the strong correlation between the power consumption of the user and meteorological, economic and other social factors, it is necessary to consider load prediction using external data, wherein the external data may include: the weather data, economic data, social information, such as one or more of weather data, holiday information, public traffic flow conditions and/or regional production total, are selected according to different electricity usage scenarios.
Here, the meteorological information includes but is not limited to temperature, humidity, light and wind speed, etc., the economic data includes but is not limited to financial index, macroscopic financial index, etc., and the social information includes but is not limited to whether it is weekend, whether it is legal for holiday, whether there is significant social activity, etc.
In order to ensure the correct mapping relationship between the electricity consumption data and the external data, the electricity consumption data and the external data need to form a one-to-one mapping data structure.
For example, the historical electricity consumption data of each individual in the sample cluster corresponding to the cluster to be predicted and the external data in the same period may be the following cases:
the first power utilization scene:
the distributed power consumption data is the power consumption load of 873 users in a certain area from 6 months to 8 months, and the external data is meteorological data (collected at intervals of 1 hour) and holiday information. The cluster load prediction problem (prediction target) is to train a prediction model to predict the daily electric quantity of the cluster load (total load of the cluster to be predicted) formed by 873 users in each day of 8 months according to the actual data of 6 and 7 months (the actual data of the sample cluster corresponding to the cluster to be predicted).
The second power utilization scene:
the distributed electricity consumption data is the electricity consumption of 2004-2015 years of autonomous regions (6) in provinces and cities of the whole country, and the external data is the total production value of the regions in 2004-2015 years. The cluster load prediction problem (prediction target) is to predict the national annual power consumption (cluster total load to be predicted) in 2013 to 2015 based on the 2004 to 2012 distributed power consumption data (actual data of the sample cluster) and 2004 to 2015 external data.
The third power utilization scenario:
the distributed electricity consumption data is an electricity load curve of each business area in a certain city, the external data is the traffic condition (public transport, subway and taxi) in the coverage area of each business area (1 hour precision), and the weather information (air temperature, rainfall and 1 hour precision) of the position of each business area. The cluster load prediction problem (prediction target) is to predict the electricity load curve of a commercial district in the whole city based on external data of a future day through training of historical data.
Step S102, learning and fitting the historical electricity consumption data of the individual and external data in the same period through a machine learning algorithm to form an electricity consumption load prediction model of the individual;
after obtaining the historical electricity consumption data of each individual in the sample cluster corresponding to the cluster to be predicted and the external data in the same period, a mapping relation between the historical electricity consumption data of the individual and the external data (meteorological data, holiday information and the like) is formed in a statistical table or the like, and a data association relation between the historical electricity consumption data and the external data is obtained.
And learning and fitting the data association relation between the historical electricity utilization data of the individual and the external data in the same period through a machine learning algorithm to form an individual electricity utilization load prediction model.
The machine learning algorithm can adopt an artificial neural network or a linear regression algorithm, and the obtained power load prediction model of the individual can be a neural network model or a linear regression model.
The structure of an artificial neural network can contain three layers of neurons, an input layer, a hidden layer, and an output layer, which can fit a variety of complex data relationships.
And S103, acquiring external data of the cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value.
Specifically, firstly, external data of a cluster to be predicted are obtained, then, the total load of the cluster to be predicted is predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and a cluster load prediction value is output.
The total load of the cluster to be predicted may be all electricity consumption data of the cluster to be predicted in a certain period of time in the future relative to the sample cluster.
Taking the first power utilization scenario as an example, for example, the currently counted distributed power utilization data is the power utilization load of 873 users in a certain area from 6 months to 8 months, the actually acquired historical power utilization load of 873 users in a certain area from 6 months to 7 months serving as a sample cluster is obtained, and the external data is meteorological data (collected at intervals of 1 hour) and holiday information. The total load of the cluster to be predicted is the daily electricity consumption of the cluster load formed by 873 users in each day of 8 months.
After the external data of the cluster to be predicted are obtained, the external data of the cluster to be predicted are input into the individual power load prediction model, the total load of the cluster to be predicted is predicted, and the cluster load prediction value is output.
Taking the first power utilization scenario as an example, according to actual data (historical power utilization data) of months 6 and 7, a prediction model is trained to predict daily power consumption of the cluster load formed by 873 users of each day of month 8 (namely, the total load of the cluster to be predicted), and finally, a predicted value of the daily power consumption of the cluster load formed by 873 users of each day of month 8 is output.
That is, the electricity consumption data of 6 and 7 months and the external data are input into the training model, the functional relationship between the electricity consumption and the external data is established, the individual electricity load prediction model is generated, and when the electricity consumption of 8 months is predicted, the external data of 8 months only needs to be input into the individual electricity load prediction model. In actual use, for example, if the electricity consumption data of the tomorrow needs to be predicted, the tomorrow weather data is acquired first, and the weather data is input to obtain the electricity consumption data of the tomorrow.
For another example, in the second electricity utilization scenario, the historical electricity utilization data of an individual is the electricity consumption of 2004 to 2012 of the municipality autonomous region (6) in the provinces of the whole country, and the external data of the same period is the total production value (GDP) of the region in 2004 to 2015. The national annual power consumption in 2013-2015 can be predicted based on the 2004-2012 distributed power consumption data and 2004-2015 external data.
Inputting the 2004-2012 electricity consumption data and the GDP into a training model, establishing a functional relation between the electricity consumption and the GDP, and only inputting the GDP data of 2013 into the model when predicting 2013. In actual use, if the electricity consumption in 2019 is predicted, the GDP in 2019 needs to be known, and the value can be provided from the outside.
According to the scheme, historical electricity consumption data of each individual in the sample cluster corresponding to the cluster to be predicted and external data of the same period are obtained; learning and fitting individual historical electricity utilization data and external data in the same period through a machine learning algorithm to form an individual electricity utilization load prediction model; acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted through an individual power load prediction model based on the external data of the cluster to be predicted, outputting a cluster load prediction value, realizing quick and accurate judgment of abnormal users, and providing data support for power services such as power scheduling and power generation planning; in addition, the total load of the cluster to be predicted can be predicted by combining an individual power load prediction model and an error correction model. Compared with the prior art, the large task of cluster load prediction is decomposed into a load prediction subtask for a large number of individuals, refined measurement data in the power system are fully utilized, so that the microscopic research and analysis on the individual data provide help for macroscopic load prediction, more measurement data are considered in the load prediction, the refined degree of the cluster load prediction is improved by utilizing refined influence factors, an individual power utilization load prediction model is built through a machine learning algorithm, and the load change trend is comprehensively considered by combining the individual power utilization data and external information such as weather and holidays, so that the prediction effect of the cluster load is improved; in addition, the load forecasting speed is accelerated by utilizing the machine learning algorithm, and the short-term load forecasting can be rapidly updated.
As shown in fig. 2, a second embodiment of the present invention provides a cluster load prediction method, based on the embodiment shown in fig. 1, in step S103, before obtaining external data of a cluster to be predicted, predicting a total load of the cluster to be predicted based on the external data of the cluster to be predicted and using an electricity load prediction model of the individual, and outputting a predicted value of the cluster load, the method further includes:
step S1031, creating an error correction model between the individual power load and the cluster total load;
in step S103, predicting the total load of the cluster to be predicted by using the power load prediction model of the individual, and outputting the cluster load prediction value includes:
and S1032, acquiring external data of the cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and combining the power load prediction model and the error correction model of the individual, and outputting a cluster load prediction value.
Because the load prediction of the individual users has errors and part of user data is removed in the data preprocessing process, the mapping relationship between each user and the cluster load needs to be established. In this embodiment, a machine learning method is adopted to fit the relationship between the individual data of a plurality of users and the cluster load, and an error correction model between the individual power load and the cluster total load is created.
As an implementation manner, the following scheme may be specifically adopted to create the error correction model between the individual power load and the cluster total load:
acquiring historical total load of a cluster to which the individual belongs in the same period as the historical electricity consumption data of each individual, and taking the historical total load as the total load of the sample cluster corresponding to the cluster to be predicted;
predicting the power load of the individuals in the sample cluster corresponding to the cluster to be predicted based on the power load prediction model of the individuals to obtain the power load prediction value of each individual;
accumulating the predicted value of the power load of each individual to obtain the predicted accumulated value of the load of the individual;
and analyzing the data association relation between the load prediction accumulated value of the individual and the total load of the sample cluster, and performing error fitting to form an error correction model between the power load of the individual and the total load of the cluster.
The total load of the sample cluster is statistical historical electricity consumption data of all individuals of the sample cluster corresponding to the cluster to be predicted, and the total load of the sample cluster is centralized data relative to distributed electricity consumption data of the individuals, and is shown in fig. 5.
The error correction model may be a linear regression model or an extreme learning machine model.
After the power load prediction model and the error correction model of the individual are obtained, the overall load of the cluster to be predicted is predicted by combining the power load prediction model and the error correction model of the individual, and the cluster load prediction value is output.
Specifically, the total load of the cluster to be predicted may be all electricity consumption data of the cluster to be predicted within a certain period of time in the future with respect to the sample cluster.
Taking the first power utilization scenario as an example, for example, the currently counted distributed power utilization data is the power utilization load of 873 users in a certain area from 6 months to 8 months, the actually acquired historical power utilization load of 873 users in a certain area from 6 months to 7 months serving as a sample cluster is obtained, and the external data is meteorological data (collected at intervals of 1 hour) and holiday information. The total cluster load to be predicted is the daily cluster load electricity consumption formed by 873 users in 8 months and each day, and a prediction model and an error correction model are trained to predict the daily cluster load electricity consumption formed by 873 users in 8 months and each day (namely the total cluster load to be predicted) according to the actual data (historical electricity consumption data) in 6 and 7 months.
According to the scheme, historical electricity utilization data of each individual in the cluster to be predicted and external data in the same period are obtained; learning and fitting the data association relation between the historical electricity utilization data of the individual and the external data in the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual; establishing an error correction model between the individual power load and the cluster total load; the method comprises the steps of obtaining external data of a cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and combining an individual power load prediction model and an error correction model, outputting a cluster load prediction value, realizing rapid and accurate judgment of abnormal users, and providing data support for power services such as power scheduling and power generation planning. Compared with the prior art, the large task of cluster load prediction is decomposed into load prediction subtasks for a large number of individuals, refined measurement data in the power system are fully utilized, so that the microscopic research and analysis on individual data provide help for macroscopic load prediction, more measurement data are considered in the load prediction, the refinement degree of the cluster load prediction is improved by utilizing refined influence factors, data errors between the individuals and the cluster loads are mined through a machine learning algorithm, the load change trend is comprehensively considered by combining the individual power consumption data and external information such as weather and holidays, and the prediction effect of the cluster loads is improved; in addition, the load forecasting speed is accelerated by utilizing the machine learning algorithm, and the short-term load forecasting can be rapidly updated.
As shown in fig. 3, a third embodiment of the present invention provides a cluster load prediction method, which is based on the embodiment shown in fig. 2, and in the step S101: after obtaining the historical electricity consumption data of each individual in the sample cluster corresponding to the cluster to be predicted and the external data in the same period, the method further comprises the following steps:
step S1011, preprocessing the historical electricity utilization data of the individual and the external data in the same period;
the step S102: learning and fitting the historical electricity utilization data of the individual and the external data in the same period through a machine learning algorithm to form an electricity load prediction model of the individual, wherein the electricity load prediction model comprises the following steps:
and step S1021, learning and fitting the preprocessed individual historical electricity consumption data and the external data in the same period through a machine learning algorithm to form an individual electricity consumption load prediction model.
In order to prevent the influence of such error information on load prediction, the original data needs to be removed or reprocessed by a data preprocessing method. Specifically, the method comprises the steps of checking abnormal values, checking data formats, inquiring data vacancy and the like, and finally ensuring the integrity, authenticity and availability of the power utilization data.
Therefore, the individual historical electricity utilization data and the external data are preprocessed mainly by processing abnormal conditions such as data loss, data errors and the like to form normalized electricity utilization data, so that accurate data are provided for subsequent load prediction, and the refinement degree and the prediction effect of cluster load prediction are improved.
The preprocessing method may be mean smoothing, adjusting time scales of the power consumption data and the external data to the same level, quantizing some external data, normalizing the external data to eliminate the influence of the absolute value, and the like.
One of the historical electricity consumption data and the external data may be optionally preprocessed as required.
Further, after a cluster load predicted value to be predicted is obtained, the cluster load predicted value can be output to an electric power service platform to provide data support for the electric power service platform; or detecting abnormal electricity utilization behaviors based on the cluster load predicted value.
According to the scheme, the historical electricity utilization data of each individual in the cluster to be predicted and the external data in the same period are obtained, and the historical electricity utilization data of the individual and the external data in the same period are preprocessed; learning and fitting the data association relation between the historical electricity utilization data of the individual and the external data in the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual; establishing an error correction model between the individual power load and the cluster total load; the method comprises the steps of obtaining external data of a cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and combining an individual power load prediction model and an error correction model, outputting a cluster load prediction value, realizing rapid and accurate judgment of abnormal users, and providing data support for power services such as power scheduling and power generation planning. Compared with the prior art, the large task of cluster load prediction is decomposed into load prediction subtasks for a large number of individuals, refined measurement data in the power system are fully utilized, so that the microscopic research and analysis on individual data provide help for macroscopic load prediction, more measurement data are considered in the load prediction, the refinement degree of the cluster load prediction is improved by utilizing refined influence factors, data errors between the individuals and the cluster loads are mined through a machine learning algorithm, the load change trend is comprehensively considered by combining the individual power consumption data and external information such as weather and holidays, and the prediction effect of the cluster loads is improved; in addition, the load forecasting speed is accelerated by utilizing the machine learning algorithm, and the short-term load forecasting can be rapidly updated.
Fig. 4 shows a schematic block diagram of cluster load prediction according to this embodiment.
The cluster load prediction flow according to this embodiment is shown in fig. 5. The main process comprises the following steps:
1. acquiring historical power consumption data of each individual in a sample cluster corresponding to a cluster to be predicted, processing abnormal conditions such as data loss, data error and the like to form normalized power consumption data, and storing the normalized power consumption data in a load database;
2. acquiring external data such as meteorological data, economic data, holiday information and the like of individual power utilization data in the same period in a sample cluster corresponding to a cluster to be predicted, storing the external data into an external information database, and forming a mapping relation between the power utilization data and the external data;
3. learning and fitting the data association relation between the individual power consumption data and external data through a machine learning algorithm to form an individual power consumption load prediction model;
4. respectively carrying out load prediction on each user according to the individual power load prediction model in the step 3 to obtain an individual load prediction accumulated value; and analyzing the data association relation between the load prediction accumulated value of the individual and the cluster load (cluster load true value), and performing error fitting to form an error correction model between the individual power consumption and the overall load.
5. And (4) predicting the cluster load data to be predicted by combining the load prediction model and the error correction model formed in the step (3) and the step (4), and finally outputting the load predicted value of the cluster user to provide data support for power services such as power dispatching and power generation planning.
Referring to fig. 6, fig. 6 is a schematic diagram illustrating comparison between load prediction and actual measurement of the residential users in the cluster load prediction method according to the embodiment of the present invention.
Referring to fig. 7, fig. 7 is a schematic diagram illustrating a comparison between prediction accuracy in test data and other methods in the cluster load prediction method according to the embodiment of the present invention.
Compared with the prior art, the embodiment of the invention has the following advantages:
1. by improving the individual load prediction precision, more measurement data are considered in the load prediction, and the prediction precision of the cluster load is improved by utilizing refined influence factors;
2. data errors between the individual and the cluster loads are mined through a machine learning algorithm, so that the prediction effect of the cluster loads is improved;
3. the machine learning algorithm is utilized to accelerate the load forecasting speed, and the short-term load forecasting can be rapidly updated.
The embodiments of the invention are illustrated in detail below by way of examples:
example 1
The distributed power consumption data is the load (every 5 minutes acquisition interval) of 873 households in a certain area from 6 months to 8 months in a certain area, the external data is meteorological data (1 hour acquisition interval) and holiday information, and the prediction target is that a prediction model is trained to predict the cluster load daily power consumption formed by 873 households in 8 months according to the actual data of 6 months and 7 months.
First, data preprocessing
1) Converting the electricity consumption data of a single user to obtain the daily electricity consumption of each user;
2) and performing mean value smoothing on the abnormal value of the power consumption, wherein the formula is as follows:
Figure BDA0001864746060000141
in the above formula, the abnormal value is the nth day; l isnThe power consumption of the nth day; m is a value day range with smooth mean value, and is set as a positive integer according to the use requirement; l isiThe used amount on the i-th day.
3) And processing the weather information according to the daily mean value to form daily weather data including temperature, humidity, rainfall and air pressure.
Secondly, training individual load prediction algorithm
According to the data acquired in the step one, taking the data of the user 242 in 6 months and 4 days as an example, the structure is as follows:
user number Electric power consumption Air temperature Humidity Rainfall amount Air pressure Whether to save holidays
242 4.31 24 68 0 1008.2 Whether or not
The content of the data is more, and is not listed here. According to the data organized according to the format, an individual load prediction is carried out, an artificial neural network is adopted to construct an individual load prediction model in the embodiment, the structure of the artificial neural network is shown in fig. 8 and comprises three layers of neurons which are respectively an input layer, a hidden layer and an output layer, the structure can fit various complex data relationships, the test result is shown in fig. 9, errors of training, verification and testing are respectively compared, and the prediction error is mainly concentrated to about +/-5%.
Third, correcting load error of cluster user
And B, respectively predicting the load of each user according to the individual load prediction method in the step B, and counting 873 users. The prediction results are accumulated and compared with the actual load power consumption of the cluster, as shown in fig. 10, the load of 15 days is predicted, and the error is about 8.3%. In order to correct errors, a multiple linear regression model is adopted to fit data to construct an error correction model, and the formula is as follows:
Figure BDA0001864746060000151
in the formula, LtotalTotal amount of electricity consumed for real cluster load, kiIs a regression coefficient, LiAnd b is a regression deviation coefficient.
The result of the error-corrected data is shown in fig. 11, and the relative error is reduced to 1.2%.
And then, predicting the cluster load to be predicted according to the load prediction model and the error correction model, and outputting a cluster load predicted value.
Compared with the prior art, the embodiment of the invention improves the prediction precision of the cluster load.
Example 2
The distributed power consumption data is 2004-2015 annual power consumption of autonomous regions (6) in provinces and cities of the whole country, the external data is 2004-2015 total production value, and the data is shown in the following table 1. The load prediction problem is that the national annual power consumption in 2013-2015 is predicted based on 2004-2012 distributed power consumption data and 2004-2015 external data.
TABLE 1 Total electric power consumption and production
Figure BDA0001864746060000152
Figure BDA0001864746060000161
First, data preprocessing
The data source of the embodiment is regular, the data is processed, the data is real and reliable, and the usability is high, so that the preprocessing work is not needed in the embodiment.
Secondly, training individual load prediction algorithm
The embodiment belongs to load prediction in a long time period, and the data sample is small, so a load prediction model is constructed by analyzing the data relation between the annual power consumption and the total regional production value of each provincial administrative unit through linear regression, and the regression formula is as follows:
L＝kM+b
in the formula, L is annual power consumption, k is a regression coefficient, and b is a regression deviation coefficient. The data regression prediction effect is shown in fig. 12, where M is the total area production value, the correlation between the power consumption and the GDP is approximately linear, and there is a constant term for adjusting the deviation.
Third, correcting load error of cluster user
In the embodiment, the distributed data only selects partial information among the cluster data, so that the problem of incompletely distributed data predicting the electricity load of the cluster is solved. The electricity consumption information of 6 provincial administration units is only accumulated, so that the difference is large on the whole, and therefore correlation mining is required, in the embodiment, an extreme learning machine is adopted to learn the data relationship between the two, an error correction model is constructed, and the test result is shown in fig. 13. In three years from 2013 to 2016, the relative errors of the load prediction are respectively 2.5%, 3.5% and 2.25%, and the effect is good.
And then, predicting the cluster load to be predicted according to the load prediction model and the error correction model, and outputting a cluster load predicted value.
Compared with the prior art, the embodiment of the invention improves the prediction precision of the cluster load.
Example 3
The distributed power consumption data is a power consumption load curve (5-minute precision) of each business area of a certain city, and the external data is the flow conditions (1-hour precision) of public transportation (buses, subways and rentals) in the coverage area of each business area and the weather information (air temperature, rainfall and 1-hour precision) of the position of each business area. The cluster load forecasting problem is to forecast the electricity load curve of a commercial district in the whole city based on external data of a future day through training of historical data.
First, data preprocessing
1. The time scales of the electricity consumption data and the external data need to be adjusted to the same level according to the electricity load curve required for predicting the problem. The external data was extended and averaged to 1 hour accuracy to one data per 5 minutes.
2. The extrinsic data is also normalized to eliminate the effect of the absolute value magnitude.
3. Carrying out quantization processing on public transportation to form a public traffic flow density index, wherein the formula is as follows:
Figure BDA0001864746060000171
wherein P is the public traffic flow density, NbusNumber of buses in a unit time, NmetroIs the amount of iron per unit time, NtaxiThe number of taxis in unit time and the length of unit time. The processed data structure is shown in table 2 below.
Table 2 example 3 data structure
Figure BDA0001864746060000172
II, training load prediction algorithm of individual business district
In the embodiment, the load prediction has more influence factors, including public transportation data and various meteorological indexes, and the data volume is larger, so that the load prediction of an individual business area is carried out by constructing an individual power load prediction model by adopting a neural network. Selecting the number of hidden nodes through an empirical formula:
Figure BDA0001864746060000181
in the formula, nhiddenNumber of hidden nodes, nhiddenFor inputting the number of attributes, nhiddenThe number of output results. The predicted result is shown in fig. 14, in the load curve 24 hours a day, the degree of fitting between the predicted value and the true value is high, but the error between individual time periods such as 17 hours and 20 hours is relatively large, and the number of hidden nodes of the neural network model is obtained by the above formula.
Third, correcting load error of cluster user
And according to the load curve prediction results of the single business area in the second step, accumulating the load curve prediction results, comparing the accumulated load curve with the cluster load curves of the business areas, performing error fitting through an extreme learning machine, and constructing an error correction model to obtain an accurate result, wherein the accurate result is shown in fig. 15. The load prediction relative error at 24 hours is less than 5%. The extreme learning machine algorithm selected by the embodiment has the characteristics of high training speed, high prediction precision and the like.
And then, predicting the cluster load to be predicted according to the load prediction model and the error correction model, and outputting a cluster load predicted value.
Example 4
The distributed power consumption data are power consumption data collected by a household intelligent socket of a certain power consumer, the external data are air temperature, information of holidays and the number of real-time people in a household, and the prediction target is to predict the total power consumption of the household on the day according to the power consumption data and the external data collected by the intelligent sockets.
First, data preprocessing
In the embodiment, the target of prediction is the total daily power consumption, the original data needs to be processed to form typical daily data, and if the frequency of data collected by the intelligent socket is high, the data needs to be accumulated all day long; the temperature data has large change all day, and the highest temperature of the day is selected as an attribute; the holiday information is a logical quantity, and needs to be converted into a final proportionality coefficient to participate in prediction through regular mining of historical data, and the real-time population of the family is in change, so that the two attributes of the number of working population and the number of the population of the family in constant residence can be simplified. The data format is shown in table 3 below.
Table 3 example 4 data structure
Figure BDA0001864746060000182
Figure BDA0001864746060000191
Secondly, training individual load prediction algorithm
The individual load prediction problem of the embodiment is an estimated prediction of incomplete data, and the power consumption of a plurality of intelligent sockets of a family is predicted based on external data. The method adopts a decision tree to carry out regression prediction, carries out parameter optimization through an MATLAB tool box (MATLAB is a computer programming language, is named from Matrix Laboratory, is specially used for processing computer data in a Matrix mode, integrates numerical calculation and a visual environment, is very visual, provides a large number of functions, and has more and more tool boxes to enable the application range to be more and more extensive), and constructs a load prediction model.
Third, correcting load error of cluster user
And on the basis of the prediction of the power consumption of each household intelligent socket in the step two, mining the data relation between the predicted load of the intelligent socket and the cluster load, predicting by adopting a multiple linear regression method, and constructing an error correction model, wherein the formula is as follows:
Figure BDA0001864746060000192
in the formula LtotalThe total electricity consumption of the load of the real cluster (namely the total electricity consumption of the family), n is the number of the intelligent sockets, and kiIs a regression coefficient, LiThe predicted value of the ith smart socket and b is the regression deviation coefficient, and the prediction result is shown in fig. 16. Fig. 16 shows relative errors of training, verification, and testing, where the error distribution of the prediction method used in this embodiment is ± 10%, further illustrating the effectiveness of the cluster load prediction method in this embodiment of the present invention.
Compared with the prior art, the cluster load prediction method based on the distributed power utilization data and the external information, disclosed by the embodiment of the invention, is used for predicting the microscopic load of the individual by combining the machine learning algorithm and analyzing the relation between the individual load prediction result and the overall load data of the cluster to obtain the cluster load prediction result, so that the refinement degree and the prediction effect of the cluster load prediction are improved, the load prediction speed is accelerated by utilizing the machine learning algorithm, and the quick update of the short-term load prediction can be realized.
Correspondingly, an embodiment of the present invention further provides a cluster load prediction apparatus, including:
the data acquisition module is used for acquiring historical electricity consumption data of each individual in a sample cluster corresponding to the cluster to be predicted and external data in the same period;
the load prediction model creating module is used for learning and fitting the historical electricity utilization data of the individual and the external data in the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual;
and the prediction module is used for predicting the total load of the cluster to be predicted through the power load prediction model of the individual and outputting a cluster load prediction value.
Further, the apparatus further comprises:
the error correction model creating module is used for creating an error correction model between the individual power load and the cluster total load;
the prediction module is also used for predicting the total load of the cluster to be predicted by combining the power load prediction model and the error correction model of the individual and outputting a cluster load prediction value.
In addition, an embodiment of the present invention further provides a cluster load prediction apparatus, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, the computer program when executed by the processor performing the operations of:
acquiring historical electricity consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data in the same period;
learning and fitting the historical electricity utilization data of the individual and the external data of the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual;
acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value.
Specifically, as shown in fig. 17, the cluster load prediction apparatus in this embodiment may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, a communication bus 1002. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory (e.g., a magnetic disk memory). The memory 1005 may alternatively be a storage device separate from the processor 1001.
Those skilled in the art will appreciate that the device configuration shown in fig. 17 is not intended to be limiting of the device and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
As shown in fig. 17, a memory 1005, which is a kind of computer storage medium, may include therein an operating system, a network communication module, a user interface module, and a cluster load prediction program.
In the apparatus shown in fig. 17, the network interface 1004 is mainly used for connecting to a network server and performing data communication with the network server; the user interface 1003 is mainly used for connecting a client (user side) and performing data communication with the client; and the processor 1001 may be configured to call the cluster load prediction program stored in the memory 1005 and perform the following operations:
acquiring historical electricity consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data in the same period;
learning and fitting the historical electricity utilization data of the individual and the external data of the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual;
acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value.
Further, the processor 1001 may be further configured to call a cluster load prediction program stored in the memory 1005, and perform the following operations:
establishing an error correction model between the individual power load and the cluster total load;
the steps of obtaining external data of the cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value comprise:
acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and combining with the individual power load prediction model and the error correction model, and outputting a cluster load prediction value.
Further, the processor 1001 may be further configured to call a cluster load prediction program stored in the memory 1005, and perform the following operations:
acquiring historical total load of a cluster to which the individual belongs in the same period as the historical electricity consumption data of each individual, and taking the historical total load as the total load of the sample cluster corresponding to the cluster to be predicted;
predicting the power load of the individuals in the sample cluster corresponding to the cluster to be predicted based on the power load prediction model of the individuals to obtain the power load prediction value of each individual;
accumulating the predicted value of the power load of each individual to obtain the predicted accumulated value of the load of the individual;
and analyzing the data association relation between the load prediction accumulated value of the individual and the total load of the sample cluster to form an error correction model between the power load of the individual and the total load of the cluster.
Further, the processor 1001 may be further configured to call a cluster load prediction program stored in the memory 1005, and perform the following operations:
preprocessing the historical electricity utilization data of the individual and the external data in the same period;
and learning and fitting the preprocessed individual historical electricity utilization data and the external data in the same period through a machine learning algorithm to form an individual electricity utilization load prediction model.
Further, the processor 1001 may be further configured to call a cluster load prediction program stored in the memory 1005, and perform the following operations:
outputting the cluster load predicted value to an electric power service platform to provide data support for the electric power service platform; or detecting abnormal electricity utilization behaviors based on the cluster load predicted value.
Furthermore, an embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when executed by a processor, the computer program implements the following operations:
acquiring historical electricity consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data in the same period;
learning and fitting the historical electricity utilization data of the individual and the external data of the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual;
acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value.
Further, the computer program when executed by the processor further performs the following:
establishing an error correction model between the individual power load and the cluster total load;
the steps of obtaining external data of the cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value comprise:
acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and combining with the individual power load prediction model and the error correction model, and outputting a cluster load prediction value.
For the principle of implementing cluster load prediction in this embodiment, please refer to the above embodiments, which are not described herein again.
Compared with the prior art, the cluster load prediction method, the cluster load prediction device and the storage medium provided by the embodiment of the invention have the advantages that the historical electricity consumption data of the individual and the external data in the same period are fitted through the learning of the machine learning algorithm, so that an electricity consumption load prediction model of the individual is formed; forecasting the total load of the cluster to be forecasted through an individual power load forecasting model, outputting a cluster load forecasting value, realizing quick and accurate judgment of abnormal users, and providing data support for power services such as power dispatching and power generation planning; in addition, the total load of the cluster to be predicted can be predicted by combining an individual power load prediction model and an error correction model. Compared with the prior art, the large task of cluster load prediction is decomposed into load prediction subtasks for a large number of individuals, refined measurement data in the power system are fully utilized, so that the microscopic research and analysis on individual data provide help for macroscopic load prediction, more measurement data are considered in the load prediction, the refinement degree of the cluster load prediction is improved by utilizing refined influence factors, data errors between the individuals and the cluster loads are mined through a machine learning algorithm, the load change trend is comprehensively considered by combining the individual power consumption data and external information such as weather and holidays, and the prediction effect of the cluster loads is improved; in addition, the load forecasting speed is accelerated by utilizing the machine learning algorithm, and the short-term load forecasting can be rapidly updated.
The above description is only for the preferred embodiment of the present invention and is not intended to limit the scope of the present invention, and all equivalent structures or flow transformations made by the present specification and drawings, or applied directly or indirectly to other related arts, are included in the scope of the present invention.

Claims (10)

1. A cluster load prediction method, comprising:
acquiring historical electricity consumption data of each individual in a sample cluster corresponding to a cluster to be predicted and external data in the same period;
learning and fitting the historical electricity utilization data of the individual and the external data of the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual;
acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value.
2. The method according to claim 1, wherein the step of obtaining external data of the cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and through the power load prediction model of the individual, and outputting a cluster load prediction value further comprises:
establishing an error correction model between the individual power load and the cluster total load;
the steps of obtaining external data of the cluster to be predicted, predicting the total load of the cluster to be predicted through the individual power load prediction model based on the external data of the cluster to be predicted, and outputting a cluster load prediction value comprise:
acquiring external data of a cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and combining with the individual power load prediction model and the error correction model, and outputting a cluster load prediction value.
3. The method of claim 2, wherein the step of creating an error correction model between the individual power loads and the cluster total load comprises:
acquiring historical total load of a cluster to which the individual belongs in the same period as the historical electricity consumption data of each individual, and taking the historical total load as the total load of the sample cluster corresponding to the cluster to be predicted;
predicting the power load of the individuals in the sample cluster corresponding to the cluster to be predicted based on the power load prediction model of the individuals to obtain the power load prediction value of each individual;
accumulating the predicted value of the power load of each individual to obtain the predicted accumulated value of the load of the individual;
and analyzing the data association relation between the load prediction accumulated value of the individual and the total load of the sample cluster to form an error correction model between the power load of the individual and the total load of the cluster.
4. The method according to any one of claims 1 to 3, wherein the step of obtaining the historical electricity consumption data of each individual in the sample cluster corresponding to the cluster to be predicted and the external data of the same period is followed by further comprising:
preprocessing the historical electricity utilization data of the individual and the external data in the same period;
the step of forming an individual power load prediction model by fitting the historical power consumption data of the individual to the external data of the same period through machine learning algorithm learning comprises the following steps:
and learning and fitting the preprocessed individual historical electricity utilization data and the external data in the same period through a machine learning algorithm to form an individual electricity utilization load prediction model.
5. The method of claim 4, wherein the external data comprises: one or more of meteorological data, economic data, social information; the individual power load prediction model is a neural network model or a linear regression model; the error correction model is a linear regression model or an extreme learning machine model.
6. The method according to any one of claims 1-3, further comprising:
outputting the cluster load predicted value to an electric power service platform to provide data support for the electric power service platform; or detecting abnormal electricity utilization behaviors based on the cluster load predicted value.
7. A cluster load prediction apparatus, comprising:
the data acquisition module is used for acquiring historical electricity consumption data of each individual in a sample cluster corresponding to the cluster to be predicted and external data in the same period;
the load prediction model creating module is used for learning and fitting the historical electricity utilization data of the individual and the external data in the same period through a machine learning algorithm to form an electricity utilization load prediction model of the individual;
and the prediction module is used for acquiring external data of the cluster to be predicted, predicting the total load of the cluster to be predicted based on the external data of the cluster to be predicted and outputting a cluster load prediction value through the power load prediction model of the individual.
8. The cluster load prediction device of claim 7, wherein the device further comprises:
the error correction model creating module is used for creating an error correction model between the individual power load and the cluster total load;
the prediction module is further configured to acquire external data of the cluster to be predicted, predict the total load of the cluster to be predicted based on the external data of the cluster to be predicted and combine the power load prediction model and the error correction model of the individual, and output a cluster load prediction value.
9. A cluster load prediction apparatus, comprising: memory, processor and computer program stored on the memory and executable on the processor, which computer program, when executed by the processor, carries out the steps of the cluster load prediction method according to any one of claims 1 to 6.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the cluster load prediction method according to any one of claims 1-6.
CN201811350529.0A 2018-11-14 2018-11-14 Cluster load prediction method and device and storage medium Pending CN111191811A (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201811350529.0A CN111191811A (en) 2018-11-14 2018-11-14 Cluster load prediction method and device and storage medium
PCT/CN2019/118309 WO2020098728A1 (en) 2018-11-14 2019-11-14 Cluster load prediction method and apparatus, and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811350529.0A CN111191811A (en) 2018-11-14 2018-11-14 Cluster load prediction method and device and storage medium

Publications (1)

Publication Number Publication Date
CN111191811A true CN111191811A (en) 2020-05-22

Family

ID=70707026

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811350529.0A Pending CN111191811A (en) 2018-11-14 2018-11-14 Cluster load prediction method and device and storage medium

Country Status (2)

Country Link
CN (1) CN111191811A (en)
WO (1) WO2020098728A1 (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111564848A (en) * 2020-06-09 2020-08-21 剑科云智(深圳)科技有限公司 Intelligent power dispatching method and power utilization load prediction device for micro power grid
CN112116153A (en) * 2020-09-18 2020-12-22 上海电力大学 Park multivariate load joint prediction method for coupling Copula and stacked LSTM network
CN112232886A (en) * 2020-10-30 2021-01-15 南方电网能源发展研究院有限责任公司 Electricity price probability prediction method, system, computer equipment and storage medium
CN112633738A (en) * 2020-12-30 2021-04-09 安徽广志电气有限公司 Distribution control cabinet terminal load distribution method
CN113131476A (en) * 2021-04-28 2021-07-16 南方电网科学研究院有限责任公司 Power load prediction method
CN113222290A (en) * 2021-06-03 2021-08-06 玲珑集团有限公司 Power load collaborative optimization method for air compressor and internal mixer cluster
CN113360270A (en) * 2021-06-30 2021-09-07 杭州数梦工场科技有限公司 Data cleaning task processing method and device
CN113592528A (en) * 2021-06-22 2021-11-02 国网河北省电力有限公司营销服务中心 Baseline load estimation method and device and terminal equipment
CN116956198A (en) * 2023-09-18 2023-10-27 湖南迈格瑞特物联科技有限公司 Intelligent electricity consumption data analysis method and system based on Internet of things

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104376371A (en) * 2014-10-31 2015-02-25 国家电网公司 Distribution network layering load forecasting method based on topology
CN104881706A (en) * 2014-12-31 2015-09-02 天津弘源慧能科技有限公司 Electrical power system short-term load forecasting method based on big data technology

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102402726B (en) * 2011-11-04 2014-08-27 中国电力科学研究院 Method for predicting electric quantity of large-scale distribution network based on regional load analysis
CN102982387A (en) * 2012-10-18 2013-03-20 安徽工程大学 Method for predicting short-term power load
CA2964806C (en) * 2014-10-26 2021-04-20 Dmitriy ANICHKOV Forecasting net load in a distributed utility grid
CN107622329A (en) * 2017-09-22 2018-01-23 深圳市景程信息科技有限公司 The Methods of electric load forecasting of Memory Neural Networks in short-term is grown based on Multiple Time Scales

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104376371A (en) * 2014-10-31 2015-02-25 国家电网公司 Distribution network layering load forecasting method based on topology
CN104881706A (en) * 2014-12-31 2015-09-02 天津弘源慧能科技有限公司 Electrical power system short-term load forecasting method based on big data technology

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
杨德昌等: "面向海量用户用电数据的集成负荷预测", 《电网技术》 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111564848A (en) * 2020-06-09 2020-08-21 剑科云智(深圳)科技有限公司 Intelligent power dispatching method and power utilization load prediction device for micro power grid
CN112116153A (en) * 2020-09-18 2020-12-22 上海电力大学 Park multivariate load joint prediction method for coupling Copula and stacked LSTM network
CN112116153B (en) * 2020-09-18 2022-10-04 上海电力大学 Park multivariate load joint prediction method coupling Copula and stacked LSTM network
CN112232886A (en) * 2020-10-30 2021-01-15 南方电网能源发展研究院有限责任公司 Electricity price probability prediction method, system, computer equipment and storage medium
CN112633738A (en) * 2020-12-30 2021-04-09 安徽广志电气有限公司 Distribution control cabinet terminal load distribution method
CN113131476A (en) * 2021-04-28 2021-07-16 南方电网科学研究院有限责任公司 Power load prediction method
CN113222290A (en) * 2021-06-03 2021-08-06 玲珑集团有限公司 Power load collaborative optimization method for air compressor and internal mixer cluster
CN113592528A (en) * 2021-06-22 2021-11-02 国网河北省电力有限公司营销服务中心 Baseline load estimation method and device and terminal equipment
CN113360270A (en) * 2021-06-30 2021-09-07 杭州数梦工场科技有限公司 Data cleaning task processing method and device
CN113360270B (en) * 2021-06-30 2024-02-27 杭州数梦工场科技有限公司 Data cleaning task processing method and device
CN116956198A (en) * 2023-09-18 2023-10-27 湖南迈格瑞特物联科技有限公司 Intelligent electricity consumption data analysis method and system based on Internet of things
CN116956198B (en) * 2023-09-18 2023-12-26 湖南迈格瑞特物联科技有限公司 Intelligent electricity consumption data analysis method and system based on Internet of things

Also Published As

Publication number Publication date
WO2020098728A1 (en) 2020-05-22

Similar Documents

Publication Publication Date Title
CN111191811A (en) Cluster load prediction method and device and storage medium
CN110610280B (en) Short-term prediction method, model, device and system for power load
Dvorkin et al. Assessing flexibility requirements in power systems
US20190228329A1 (en) Prediction system and prediction method
CN105678398A (en) Power load forecasting method based on big data technology, and research and application system based on method
CN109299438B (en) Public transport facility supply level evaluation method based on network appointment data
CN110991875B (en) Platform user quality evaluation system
CN109376772B (en) Power load combination prediction method based on neural network model
JP2020501491A (en) System and method for dynamic energy storage system control
CN111612228A (en) User electricity consumption behavior analysis method based on electricity consumption information
CN113538067B (en) Inter-city network vehicle-closing demand prediction method and system based on machine learning
CN108256724B (en) Power distribution network open capacity planning method based on dynamic industry coefficient
CN113611105B (en) Urban traffic travel demand total quantity prediction method
CN116186548B (en) Power load prediction model training method and power load prediction method
WO2022021727A1 (en) Urban complex electricity consumption prediction method and apparatus, electronic device, and storage medium
CN111680841A (en) Short-term load prediction method and system based on principal component analysis and terminal equipment
Semmelmann et al. Load forecasting for energy communities: A novel LSTM-XGBoost hybrid model based on smart meter data
CN115545333A (en) Method for predicting load curve of multi-load daily-type power distribution network
CN111626543A (en) Method and device for processing power related data
Stephen et al. Non-Gaussian residual based short term load forecast adjustment for distribution feeders
Zheng et al. Short-term energy consumption prediction of electric vehicle charging station using attentional feature engineering and multi-sequence stacked Gated Recurrent Unit
CN110489893B (en) Variable weight-based bus load prediction method and system
CN111144628A (en) Distributed energy supply type cooling, heating and power load prediction model system and method
CN116245246A (en) Smart city power supply management method, system and storage medium based on Internet of things
CN113837486B (en) RNN-RBM-based distribution network feeder long-term load prediction method

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20200522