CN112465250A - Power load prediction method, power load prediction device, computer equipment and storage medium - Google Patents

Power load prediction method, power load prediction device, computer equipment and storage medium Download PDF

Info

Publication number
CN112465250A
CN112465250A CN202011423514.XA CN202011423514A CN112465250A CN 112465250 A CN112465250 A CN 112465250A CN 202011423514 A CN202011423514 A CN 202011423514A CN 112465250 A CN112465250 A CN 112465250A
Authority
CN
China
Prior art keywords
power load
modeling
load data
target
prediction model
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
CN202011423514.XA
Other languages
Chinese (zh)
Inventor
胡子珩
李艳
张华赢
陶骏
潘天红
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.)
Shenzhen Power Supply Bureau Co Ltd
Original Assignee
Shenzhen Power Supply Bureau 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 Shenzhen Power Supply Bureau Co Ltd filed Critical Shenzhen Power Supply Bureau Co Ltd
Priority to CN202011423514.XA priority Critical patent/CN112465250A/en
Publication of CN112465250A publication Critical patent/CN112465250A/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
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/21—Design, administration or maintenance of databases
    • G06F16/211—Schema design and management
    • G06F16/212—Schema design and management with details for data modelling support
    • 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
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • 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

Abstract

The application relates to a power load prediction method, a power load prediction device, computer equipment and a storage medium, which are applied to the technical field of power systems. The method comprises the following steps: acquiring m pieces of historical load data in a preset time period before the current moment, wherein m is an integer greater than 1; obtaining a plurality of sample sets from the m historical load data, each sample set comprising lmThe historical load data in each sample set have the relevance larger than a preset relevance threshold value lmIs an integer greater than 1; calculating parameters of a target power load prediction model based on the plurality of sample sets; parameters based on the target power load prediction modelAnd predicting the power load data after the current moment by the target power load prediction model. By adopting the method, the target power load prediction model can be constructed in real time, and the time sequence of the power load is considered, so that the power load data obtained by prediction is more accurate.

Description

Power load prediction method, power load prediction device, computer equipment and storage medium
Technical Field
The present application relates to the field of power system technologies, and in particular, to a power load prediction method, apparatus, computer device, and storage medium.
Background
With the development of science and technology, the development of power systems is more and more perfect. The power load prediction plays an increasingly important role in the development process of a power system, and the result of the power load prediction is an important basis for power system scheduling, planning and equipment maintenance. The accurate power load prediction can plan the start and stop of the generator set, optimize the load distribution of photoelectricity, wind power, hydropower and thermal power, reduce energy consumption, and has great significance for economic sustainable development, social benefit and environmental protection. Therefore, how to accurately predict the power load becomes an increasingly important issue.
In the conventional technology, it is generally required to process historical power load data, extract features, train a machine learning model by using the historical power load data after the features are extracted, and predict the power load data by using the trained machine learning model.
However, since the machine learning model training process is complicated, and the machine learning model is trained only once using the historical load data, the dynamic performance of the power load data is not considered, and thus, the data of the power load prediction is inaccurate.
Disclosure of Invention
In view of the above, it is necessary to provide a power load prediction method, a power load prediction apparatus, a computer device, and a storage medium, which can capture dynamic changes of a power system in time and effectively predict power load data.
In a first aspect, a power load prediction method is provided, and the method includes: acquiring m pieces of historical load data m in a preset time period before the current timeObtaining a modeling sample set from m historical load data for integers larger than 1, wherein the modeling sample set comprises a plurality of modeling samples, and each modeling sample comprises lmThe historical load data in each modeling sample has the relevance larger than a preset relevance threshold value lmIs an integer greater than 1; calculating parameters of a target power load prediction model based on the modeling sample set; and predicting the power load data after the current moment based on the parameters of the target power load prediction model and the target power load prediction model.
In one embodiment, obtaining a set of modeling samples from m historical load data includes: calculating the time series length l of the target modeling samplemLength of time series lmThe maximum number of the time-continuous historical load data with correlation in the m pieces of historical load data is represented; based on the length of the time series lmAnd acquiring a plurality of modeling samples from the m pieces of historical load data to construct a modeling sample set.
In one embodiment, the time series length l of the target sample set is calculatedmThe method comprises the following steps: based on the partial autocorrelation coefficients, sequentially calculating the correlation degree between the current load data and the historical load data which is continuous in time before the current time from the load data corresponding to the current time in the m pieces of historical load data; when the correlation degree is smaller than a preset correlation degree value, determining the time sequence length l of the target sample setm。
In one embodiment based on the length of the time series lmObtaining a plurality of modeling samples from the m historical load data, comprising: determining a candidate modeling sample set based on the m historical load data, wherein the candidate modeling sample set comprises a plurality of candidate modeling samples; respectively calculating similarity indexes of the target sample and a plurality of candidate modeling samples in the candidate modeling sample set; and screening candidate modeling samples with similarity indexes larger than preset similarity indexes with the target sample from the candidate modeling sample set based on the similarity indexes, so as to obtain a plurality of modeling samples.
In one of the embodimentsThe target power load prediction model is as follows:
Figure BDA0002823632720000021
in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000022
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) Is historical load data; lpTo predict the step size, if l p1, i.e. single step prediction,/p>1 is the multi-step prediction, and the prediction method,
Figure BDA0002823632720000023
the output of the model is predicted for the target electrical load.
In one embodiment, calculating parameters of the target power load prediction model based on the modeling sample set includes: based on the formula
Figure BDA0002823632720000024
Calculating parameters of a target power load prediction model;
in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000025
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) For historical load data, k is the number of modeling samples in the set of modeling samples, wi＝1-D(φi,φt) The weight of the modeled sample, D (phi), determined for the similarity indexi,φt) Modeling a sample phi for the targettAnd a modeling sample phiiSimilarity index between, yiFor the modeling sample
Figure BDA0002823632720000031
The corresponding power load value.
In one embodiment, predicting the power load book data after the current time based on the parameters of the target power load prediction model and the target power load prediction model includes: acquiring a previous momentm(ii) historical load data; constructing a target power load prediction model based on parameters of the target power load prediction model; the current time is the previous onemThe historical load data of each time series is input into a target power load prediction model, and power load data after the predicted current time is obtained.
In a second aspect, there is provided an electrical load prediction apparatus, comprising:
the first acquisition module is used for acquiring m pieces of historical load data in a preset time period before the current moment, wherein m is an integer greater than 1;
a second obtaining module, configured to obtain a modeling sample set from the m pieces of historical load data, where the modeling sample set includes a plurality of modeling samples, and each modeling sample includes lmThe historical load data in each modeling sample has the relevance larger than a preset relevance threshold value lmIs an integer greater than 1;
the calculation module is used for calculating parameters of the target power load prediction model based on the modeling sample set; (ii) a
And the prediction module is used for predicting the power load data after the current moment based on the parameters of the target power load prediction model and the target power load prediction model.
In one embodiment, the second obtaining module includes a calculating unit and a obtaining unit, wherein:
a calculation unit for calculating the time-series length l of the calculation target modeling samplemLength of time series lmThe maximum number of the time-continuous historical load data with correlation in the m pieces of historical load data is represented;
an acquisition unit for acquiring a time-series length lmAnd acquiring a plurality of modeling samples from the m pieces of historical load data to construct a modeling sample set.
In one embodiment, the calculating unit is specifically configured to: based on the partial autocorrelation coefficients, sequentially calculating the correlation degree between the current load data and the historical load data which is continuous in time before the current time from the load data corresponding to the current time in the m pieces of historical load data; when the correlation degree is smaller than a preset correlation degree value, determining the time sequence length l of the target sample setm。
In one embodiment, the obtaining unit is specifically configured to: determining a candidate modeling sample set based on the m historical load data, wherein the candidate modeling sample set comprises a plurality of candidate modeling samples; respectively calculating similarity indexes of the target sample and a plurality of candidate modeling samples in the candidate modeling sample set; and screening candidate modeling samples with similarity indexes larger than preset similarity indexes with the target sample from the candidate modeling sample set based on the similarity indexes, so as to obtain a plurality of modeling samples.
In one embodiment, the calculating module is specifically configured to: the method for constructing the target power load prediction model comprises the following steps:
Figure BDA0002823632720000041
in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000042
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) Is historical load data; lpTo predict the step size, if l p1, i.e. single step prediction,/p>1 is the multi-step prediction, and the prediction method,
Figure BDA0002823632720000043
the output of the model is predicted for the target electrical load.
In one embodiment, the calculating module is specifically configured to: based on the formula
Figure BDA0002823632720000044
Eyes of calculationParameters of a standard power load prediction model; in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000045
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) For historical load data, k is the number of modeling samples in the set of modeling samples, wi＝1-D(φi,φt) The weight of the modeled sample, D (phi), determined for the similarity indexi,φt) Modeling a sample phi for the targettAnd a modeling sample phiiSimilarity index between, yiFor the modeling sample
Figure BDA0002823632720000046
The corresponding power load value.
In one embodiment, the prediction module is specifically configured to: acquiring a previous momentm(ii) historical load data; constructing a target power load prediction model based on parameters of the target power load prediction model; the current time is the previous onemThe historical load data of each time series is input into a target power load prediction model, and power load data after the predicted current time is obtained.
In a third aspect, there is provided a computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the power load prediction method according to any one of the first aspect described above when executing the computer program.
In a fourth aspect, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the power load prediction method according to any one of the first aspect.
According to the power load prediction method, the power load prediction device, the computer equipment and the storage medium, m pieces of historical load data in a preset time period before the current time are obtained, wherein m is an integer greater than 1; obtaining a modeling sample set from m pieces of historical load data, wherein the modeling sample set comprises a plurality of modelsModel samples, each model sample comprisingmThe correlation of the historical load data in each modeling sample is larger than a preset correlation threshold value,/, in each modeling samplemIs an integer greater than 1; calculating parameters of a target power load prediction model based on the modeling sample set; and predicting the power load data after the current moment based on the parameters of the target power load prediction model and the target power load prediction model. According to the method, the target power load prediction model is constructed according to the historical load data, so that the power load data after the current moment can be predicted.
Drawings
FIG. 1 is a diagram of an exemplary embodiment of a power load prediction method;
FIG. 2 is a flow diagram illustrating a method for predicting a power load according to one embodiment;
FIG. 3 is a diagram illustrating the acquisition of user power usage in one embodiment;
FIG. 4 is a flow chart illustrating a method for predicting a power load according to another embodiment;
FIG. 5 is a flow chart illustrating a method for predicting a power load according to another embodiment;
FIG. 6 is a flow chart illustrating a method for predicting a power load according to another embodiment;
FIG. 7 is a flow chart illustrating a method for predicting a power load according to another embodiment;
FIG. 8 is a diagram illustrating the results of a single step prediction in power load prediction in another embodiment;
FIG. 9 is a flow chart illustrating a method for predicting a power load according to another embodiment;
FIG. 10 is a flow chart illustrating a method for predicting a power load according to another embodiment;
FIG. 11 is a block diagram showing the structure of a power load prediction apparatus according to an embodiment;
fig. 12 is a block diagram showing the structure of a power load prediction apparatus according to another embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The power load prediction method provided by the application can be applied to the application environment shown in fig. 1. The application environment provides a computer device, which may be a computer device, the internal structure of which may be as shown in fig. 1. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing power load prediction data. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a power load prediction method. When the computer device is a terminal, the display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, a key, a trackball or a touch pad arranged on a casing of the computer device, or an external keyboard, a touch pad or a mouse.
Those skilled in the art will appreciate that the architecture shown in fig. 1 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, as shown in fig. 2, there is provided a power load prediction method, which is described by taking the method as an example applied to the computer device in fig. 1, and includes the following steps:
step 201, a computer device obtains m pieces of historical load data in a preset time period before the current time. Wherein m is an integer greater than 1.
In the embodiment of the present application, the historical load data may be a sum of power consumption data for each user before the current time. As shown in fig. 3, the manner of acquiring the electricity consumption data of each user by the computer device may be: the computer equipment can firstly send a request for acquiring power consumption data to the electric meters of all users according to the standard protocol of the power industry, the electric meters receive the request for acquiring the power consumption data sent by the computer equipment, and the power consumption data are converted into infrared light and sent to the computer equipment according to the request for acquiring the power consumption data sent by the computer equipment. The computer equipment acquires infrared light emitted by the electric meter of each user and analyzes the infrared light, so that power consumption data on the electric meter of each user is acquired.
In this embodiment of the present application, the interval time for the computer device to acquire the power consumption data sent by each user electric meter may be 15 minutes, half an hour, or one hour.
In the embodiment of the application, after the computer device obtains the power consumption data of each user, the power consumption data of each user can be accumulated, so that the historical load data of the sum of the power consumption data of all the users is obtained.
At step 202, the computer device obtains a set of modeling samples from the m historical load data.
Wherein, the modeling sample set comprises a plurality of modeling samples, each modeling sample comprises lmThe historical load data in each modeling sample has the relevance larger than a preset relevance threshold value lmIs an integer greater than 1.
In the embodiment of the application, based on the method, the computer device may obtain m pieces of historical load data before the current time in the form of a historical load database. Since the daily power consumption of each user is not used, especially the difference between the holidays and the normal working days is large, so that the difference between the sum of the power consumption data of all the users is large, and the use of the historical load data is not facilitated, the historical load data needs to be normalized. In this embodiment of the present application, optionally, the normalization processing mode may be:
Figure BDA0002823632720000071
in the formula, sorg(i) Is the original historical load data, t is the current time, i is the sample number,
Figure BDA0002823632720000072
is the average of all the historical load data,
Figure BDA0002823632720000073
and s (i) is the standard deviation of all the historical load data, and is the historical load data after normalization processing.
And screening a modeling sample set from the m pieces of historical load data after the normalization processing, and using the modeling sample set to construct a target power load prediction model. Wherein, the modeling sample set comprises a plurality of modeling samples, each modeling sample comprises normalized lmThe historical load data respectively calculates the correlation of the historical load data in each modeling sample, and according to the correlation calculation result, the correlation of the historical load data in each modeling sample screened out from the correlation calculation result is greater than a preset correlation threshold value lmFor integers greater than 1, the preset correlation threshold may be set according to the actual data correlation.
Step 203, the computer device calculates parameters of the target power load prediction model based on the set of modeling samples.
In this embodiment, optionally, the computer device may determine, based on the obtained modeling sample set, next historical load data corresponding to each modeling sample for each modeling sample in the modeling sample set, and use the found next historical load data corresponding to each modeling sample as an output of the target power load prediction model, and use each sample set as an input of the target power load prediction model, so as to determine parameters of multiple sets of target power load prediction models according to multiple sets of inputs and outputs of the target power load prediction model, and may perform optimization processing based on a least square method on the parameters of the multiple sets of target power load prediction models, so as to finally obtain the parameters of the target power load prediction model.
In step 204, the computer device predicts the power load data after the current time based on the parameters of the target power load prediction model and the target power load prediction model.
In an embodiment of the present application, after determining the parameters of the target power load prediction model, the computer device may construct the target power load prediction model based on the parameters of the target power load prediction model. In order to predict the power load data after the current time, the computer device may acquire historical load data before the current time with respect to an input of a target power load prediction model, and use the acquired historical load data before the current time as an input, and an output of the target power load prediction model is a prediction result of the power load data.
The power load prediction method acquires m pieces of historical load data in a preset time period before the current moment, wherein m is an integer greater than 1; obtaining a modeling sample set from m pieces of historical load data, wherein the modeling sample set comprises a plurality of modeling samples, and each modeling sample comprises lmThe correlation of the historical load data in each modeling sample is larger than a preset correlation threshold value,/, in each modeling samplemIs an integer greater than 1; calculating parameters of a target power load prediction model based on the modeling sample set; and predicting the power load data after the current moment based on the parameters of the target power load prediction model and the target power load prediction model. The method constructs the target by constructing the target based on historical load dataThe power load prediction model is standardized, so that the power load data after the current moment can be predicted, a fixed model does not need to be established, a target power load prediction model is constructed in real time, the dynamic property of the power load is considered, and the predicted power load data are more accurate.
In an alternative embodiment of the present application, as shown in fig. 4, the obtaining of the modeling sample set from the m pieces of historical load data may include the following steps:
step 401, the computer device calculates the time series length l of the target modeling samplem。
Wherein the time series length lmThe maximum number of the time-continuous historical load data with the correlation in the m pieces of historical load data is represented.
In the embodiment of the application, the computer device calculates the correlation between the historical load data closest to the current time and the next historical load data in sequence from the historical load data closest to the current time based on the acquired m pieces of historical load data. For example, assuming that a historical load data closest to the current time is a first data, a second historical load data, a third historical load data and a fourth historical load data …, calculating a correlation between the first historical load data and the second historical load data, a correlation between the first historical load data, the second historical load data and the third historical load data, a correlation between the first historical load data and the fourth historical load data and a correlation between the first historical load data and the fifth historical load data in turn from the first historical load data, … assuming that a correlation between the first historical load data and the sixth historical load data is calculated, determining that a correlation value between the first historical load data and the sixth historical load data is greater than a preset correlation threshold value according to a calculation result, and determining that the time series length of the target modeling sample is 6, and taking a historical load data set formed by the first historical load data and the sixth historical load data as the target modeling sample.
Step 402, the computer device length l based on time seriesmAnd acquiring a plurality of modeling samples from the m pieces of historical load data to construct a modeling sample set.
In the embodiment of the present application, the computer device finds the time-series length l of the target modeling sample based on the abovemFrom the remaining m-l in sequencemExtracting multiple groups of time sequence length l from historical datamAnd comparing the extracted multiple groups of time series with the length of lmHistorical load data of and target modeling samples ofmThe correlation between the historical load data and the target modeling sample is extractedmThe length of the multiple groups of time sequences with historical load data correlation larger than a preset correlation threshold is lmThe historical load data of (2) form a plurality of modeling samples, and form a modeling sample set based on the plurality of modeling samples.
In this embodiment, the time series length l of the target modeling sample is calculatedmAnd based on the time series length lmA plurality of modeling samples are obtained from the m historical load data. Therefore, a plurality of modeling samples with correlation are found, and the inaccuracy of the prediction model of the target power load caused by the fact that the sample data is irrelevant is avoided, so that the inaccuracy of the prediction of the power load data is avoided.
In an alternative embodiment of the present application, as shown in fig. 5, the time-series length l of the target sample set is calculated as described abovemMay comprise the steps of:
step 501, based on the partial autocorrelation coefficients, the computer device calculates the correlation between the current load data and the historical load data that is continuous in time before the current time in turn from the load data corresponding to the current time in the m pieces of historical load data.
In an embodiment of the present application, based on the partial autocorrelation coefficients, the computer device determines that the time series length of the target sample set is/mWherein the partial autocorrelation function is a method for describing the structural characteristics of the random process.
In bookIn the embodiment of the application, optionally, the following formula can be utilized:
Figure BDA0002823632720000101
calculating the length l of the time seriesm。
Wherein s (i) is the normalized historical load data,
Figure BDA0002823632720000102
the average value of the normalized historical load data is obtained. t is the current time, i is the sample number, RlmIs a correlation value.
In the embodiment of the application, when the correlation value RlmWhen the time sequence length is less than the preset correlation threshold, the computer equipment determines that the time sequence length of the target sample set is lm。
Step 502, when the correlation degree is smaller than the preset correlation degree value, the computer equipment determines the time sequence length l of the target modeling samplem。
In the embodiment of the present application, based on the content of the foregoing embodiment, optionally, the preset correlation threshold may be 2 σ, if R islm<2 σ, the computer device determines the time series length l of the target modeled samplem(ii) a If R islmAnd if the correlation between the historical load data currently calculated is not large, the computer needs to continuously calculate the correlation between the historical load data in sequence. Where σ is the standard deviation of the historical load data.
In the embodiment of the application, based on the partial autocorrelation coefficient, the correlation between the current load data and the historical load data before the current time is sequentially calculated from the load data corresponding to the current time in the m pieces of historical load data, and when the correlation is smaller than a preset correlation value, the time series length l of the target modeling sample is determinedm. By the method, the reasonability of the time sequence of the target modeling sample and the correlation among the historical load data in the target modeling sample can be ensured, so that the accuracy of the target power load prediction model and the reliability of the prediction result can be ensured.
In an alternative embodiment of the present application, as shown in FIG. 6, the above-mentioned length l is based on the time seriesmObtaining a plurality of modeling samples from the m historical load data may include the steps of:
step 601, the computer device determines a candidate modeling sample set based on the m historical load data.
Wherein the set of candidate modeling samples includes a plurality of candidate modeling samples.
In the embodiment of the application, the time series length l of the target modeling sample is determined from m pieces of historical load datamThereafter, the computer device may select m-l from the restmScreening a plurality of time series with the length of l from historical load datamThe candidate modeling samples of (1).
In step 602, the computer device calculates similarity indicators of the target modeling sample and a plurality of candidate modeling samples in the set of candidate modeling samples, respectively.
In the embodiment of the application, the computer device sequentially calculates the similarity indexes of the target modeling sample and each candidate modeling sample. In the embodiment of the present application, the similarity index may include a euclidean distance between the target modeling sample and the candidate modeling sample, a trend similarity index, or may be a combination of the euclidean distance between the target modeling sample and the candidate modeling sample and the trend similarity index. The similarity index between the target modeling sample and each candidate modeling sample is not specifically limited in the present application.
Optionally, in this embodiment of the present application, the computer device calculates the similarity index between the target modeling sample and each candidate modeling sample, and may use the following formula:
D(φt,φj)＝α(1-exp(-d(φt,φj)))+(1-α)δ(φt,φj)；
d(φt,φj)＝||φt-φj||2；
Figure BDA0002823632720000121
ξ(φt)＝[cos(∠(st(t-lm+1),st(t-lm))),…cos(∠(st(t),st(t-1)))]；
Figure BDA0002823632720000124
Figure BDA0002823632720000122
Figure BDA0002823632720000123
wherein d (phi)t,φj) Modeling a sample phi for an objecttWith candidate modeling samples phijOf between delta (phi) and delta (phi)t,φj) Modeling a sample phi for an objecttWith candidate modeling samples phijThe trend between them is similar to the index D (phi)t,φj) Modeling a sample phi for an objecttWith respective candidate modeling samples phijSimilarity index between them, ξ (φ)t) Modeling a sample phi for an objecttAngle vector between historical load data in (1), ξ (φ)j) Candidate modeling sample phijIs the angle vector between the historical load data in (c), cos ([ s ](s) ]t(t-i+1),st(t-lm) ()) is calendar load data(s)t(t-i +1) and st(t-lm) The included angle between the two adjacent historical load data is delta t, the interval time between two adjacent historical load data is s (t), s (t-1), … s (t-l)m) For historical load data, α is a weighting factor, and α can be determined according to actual conditions.
Step 603, based on the similarity index, the computer device screens candidate modeling samples from the candidate modeling sample set, wherein the similarity index of the candidate modeling samples with the target modeling sample is larger than a preset similarity index, and therefore a plurality of modeling samples are obtained.
In this applicationIn the embodiment, the smaller the similarity index between the target modeling sample and each candidate modeling sample is, the more similar the target modeling sample and each candidate modeling sample are. E.g. based on D (phi) in the content of the above embodimentst,φj)＝α(1-exp(-d(φt,φj)))+(1-α)δ(φt,φj) Formula, calculating D (phi) between target modeling sample and each candidate modeling samplet,φj) If D (phi)t,φj) The smaller the value, the more the candidate modeling sample phi is illustratedjSample phi of modeling with targettThe more similar.
In the embodiment of the application, the computer device screens candidate modeling samples with similarity indexes larger than a preset similarity index from a plurality of candidate modeling samples according to the similarity indexes between the target modeling sample and each candidate modeling sample to form a modeling sample set.
For example, the computer device selects a modeling sample set from the target modeling sample and each candidate modeling sample based on the formula of the similarity index between the target modeling sample and each candidate modeling sample
Figure BDA0002823632720000131
While
Figure BDA0002823632720000132
Satisfies D (phi)t,φ1)≥D(φt,φ2)≥…，
Figure BDA0002823632720000133
For the last modeling sample screened out,
Figure BDA0002823632720000134
can be determined according to actual use conditions.
In the embodiment of the application, the computer equipment determines a plurality of candidate modeling samples based on m pieces of historical load data, and respectively calculates similarity indexes of a target modeling sample and the candidate modeling samples; and screening candidate modeling samples with similarity indexes larger than preset similarity indexes with the target modeling sample from the candidate modeling samples based on the similarity indexes, thereby obtaining a plurality of modeling samples. According to the method, a plurality of candidate sets with high correlation with the target modeling sample are screened out based on the similarity indexes between the target modeling sample and the candidate modeling sample, so that the number of the sample sets is enriched, the error of constructing the target power load prediction model is reduced, the constructed target power load prediction model is more accurate, and the predicted power load data is more prepared.
In an alternative embodiment of the present application, as shown in fig. 7, the process of constructing the target power load prediction model may include the following steps:
step 701, the computer device constructs a target formula:
Figure BDA0002823632720000135
wherein the content of the first and second substances,
Figure BDA0002823632720000136
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) Is historical load data; lpTo predict the step size, if l p1, i.e. single step prediction,/p>1 is the multi-step prediction, and the prediction method,
Figure BDA0002823632720000137
the output of the model is predicted for the target electrical load.
In the embodiment of the application, the computer device determines input data and output data of each modeling sample based on the obtained plurality of modeling samples, and constructs a target power load prediction model based on the input data and the output data of each modeling sample.
For example, the modeling samples obtained by the computer device are respectively phi1,φ2,…,φtAnd predict the step length lpIf 1, find out the modeling sample phi respectively1,φ2,…,φtCorrespond to and canNext historical load data
Figure BDA0002823632720000138
Then respectively convert phi into1,φ2,…,φtAs an input to the process, the process may,
Figure BDA0002823632720000139
as output, a plurality of sets are calculated
Figure BDA00028236327200001310
FIG. 8 is a graph showing the results of the single-step prediction.
Step 702, based on formula
Figure BDA00028236327200001311
The computer device calculates parameters of a target power load prediction model.
In the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000141
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) The number of modeling samples, w, in the set of modeling samples for historical load data ki＝1-D(φi,φt) The weight of the modeled sample, D (phi), determined for the similarity indexi,φt) Modeling a sample phi for the targettAnd a modeling sample phiiSimilarity index between, yiFor the modeling sample
Figure BDA0002823632720000146
The corresponding power load value. .
In the embodiment of the application, the computer equipment obtains parameters of a plurality of groups of target power load prediction models based on calculation
Figure BDA0002823632720000142
Determining the parameters based on multiple groups of target power load prediction models by using a least square methodParameters of optimal target power load prediction model
Figure BDA0002823632720000143
Among them, the least square method (also called the least squares method) is a mathematical optimization technique. It finds the best functional match of the data by minimizing the sum of the squares of the errors. Unknown data can be easily obtained by the least square method, and the sum of squares of errors between these obtained data and actual data is minimized.
In the embodiment of the present application, the computer device constructs a target power load prediction model based on a plurality of modeling samples and output data corresponding to the plurality of modeling samples as follows:
Figure BDA0002823632720000144
and based on formulas
Figure BDA0002823632720000145
And calculating parameters of the target power load prediction model. Instead of only utilizing one group of input and output to construct the target power load prediction model, the accuracy of the constructed target power load prediction model is ensured. The method of the embodiment adopts the instant learning algorithm, utilizes the idea of point-by-point linearization, describes the nonlinear characteristic of the load of the whole power system by a plurality of local linear functions, reduces the complexity of off-line modeling, and improves the efficiency of the functions.
In an alternative embodiment of the present application, as shown in fig. 9, the predicting the power load book data after the current time based on the parameters of the target power load prediction model and the target power load prediction model may include the following steps:
step 901, the computer device obtains the current time prior to the current timemAnd (4) historical load data.
In the embodiment of the present application, the computer device obtains, based on the target power load prediction model constructed as described above, the current time prior to the current time corresponding to the target power load prediction modelmAnd (4) historical load data.
Step 902, the computer device constructs a target power load prediction model based on parameters of the target power load prediction model.
In the embodiment of the application, the computer device calculates the parameters of the target power load prediction model based on the method, and constructs the target power load prediction model based on the parameters of the target power load prediction model.
Step 903, the computer device compares the current time with the previous timemThe historical load data of each time series is input into a target power load prediction model, and power load data after the predicted current time is obtained.
In the embodiment of the present application, the computer device obtains the current time prior to the current time corresponding to the target power load prediction modelmAnd calculating the output of the target power load prediction model by using the historical load data as the input of the target power load prediction model, wherein the output of the target power load prediction model is the power load data after the predicted current time.
In the embodiment of the application, the computer equipment acquires the current time prior to the current timemAnd constructing a target power load prediction model based on the historical load data and the parameters of the target power load prediction model. The computer device will compare the current time prior tomThe historical load data is input to a target power load prediction model, and predicted power load data after the current time is obtained. According to the method, the computer equipment predicts and obtains the power load data after the current moment based on the target power load prediction model constructed in real time under the condition of considering the time sequence of the power load, and the prediction result is more accurate. And an instant learning algorithm is adopted, a prediction model of the power load does not need to be constructed offline, dynamic changes of the power system can be captured in time, the defect of poor adaptability of a fixed model is overcome, and the degree of freedom of modeling is improved.
Referring to fig. 10, a flowchart of an exemplary power load prediction method provided by an embodiment of the present application is shown, where the method may be applied to a computer device in the implementation environment shown in fig. 1. As shown in fig. 10, the method may include the steps of:
step 1001, a computer device obtains m pieces of historical load data in a preset time period before a current time, where m is an integer greater than 1.
In step 1002, based on the partial autocorrelation coefficients, the computer device calculates the correlation between the current load data and the historical load data that is continuous in time before the current time in turn from the load data corresponding to the current time in the m pieces of historical load data.
Step 1003, when the correlation degree is smaller than the preset correlation degree value, the computer equipment determines the time sequence length l of the target modeling samplem。
Based on the m historical load data, the computer device determines a set of candidate modeling samples, step 1004.
Step 1005, the computer device calculates similarity indicators of the target modeling sample and a plurality of candidate modeling samples in the candidate modeling sample set, respectively.
Step 1006, based on the similarity index, the computer device screens candidate modeling samples from the candidate modeling sample set, wherein the similarity index of the candidate modeling samples and the target modeling sample is greater than a preset similarity index, so as to obtain a plurality of modeling samples.
Step 1007, the computer device constructs a target power load prediction model as:
Figure BDA0002823632720000161
step 1008, based on the formula
Figure BDA0002823632720000162
The computer device calculates parameters of a target power load prediction model.
Step 1009, the computer device obtains l before the current timemAnd (4) historical load data.
Step 1010, the computer device constructs a target power load prediction model based on the parameters of the target power load prediction model.
Step 1011, the computer device sends the current time to the previous timemA time-continuous history loadThe data is input to a target power load prediction model to obtain predicted power load data after the current time.
It should be understood that although the various steps in the flowcharts of fig. 2, 3-7, and 9-10 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2, 3-7, and 9-10 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed alternately or at least partially with other steps or with at least some of the other steps.
In one embodiment, as shown in fig. 11, there is provided an electrical load prediction apparatus 1100, including: a first obtaining module 1101, a second obtaining module 1102, a calculating module 1103 and a predicting module 1104, wherein:
a first obtaining module 1101, configured to obtain m pieces of historical load data in a preset time period before a current time, where m is an integer greater than 1;
a second obtaining module 1102, configured to obtain a modeling sample set from the m historical load data, where the modeling sample set includes a plurality of modeling samples, and each modeling sample includes lmThe historical load data in each modeling sample has the relevance larger than a preset relevance threshold value lmIs an integer greater than 1;
a calculating module 1103, configured to calculate parameters of the target power load prediction model based on the modeling sample set;
and a prediction module 1104 for predicting the power load data after the current time based on the parameters of the target power load prediction model and the target power load prediction model.
In an alternative embodiment of the present application, as shown in fig. 12, the second obtaining module 1102 includes: a calculation unit 11021 and an acquisition unit 11022, wherein:
a calculating unit 11021 for calculating a time series length l of the target modeling samplemLength of time series lmThe maximum number of the time-continuous historical load data with the correlation in the m pieces of historical load data is represented.
An acquisition unit 11022 based on the time series length lmAnd acquiring a plurality of modeling samples from the m pieces of historical load data to construct a modeling sample set.
In an optional embodiment of the present application, the calculating unit 11021 is specifically configured to: based on the partial autocorrelation coefficients, sequentially calculating the correlation degree between the current load data and the historical load data which is continuous in time before the current time from the load data corresponding to the current time in the m pieces of historical load data; when the correlation degree is smaller than a preset correlation degree value, determining the time sequence length l of the target modeling samplem。
In an optional embodiment of the present application, the obtaining unit 11022 is specifically configured to: determining a candidate modeling sample set based on the m historical load data, wherein the candidate modeling sample set comprises a plurality of candidate modeling samples; respectively calculating similarity indexes of the target modeling sample and a plurality of candidate modeling samples in the candidate modeling sample set; and screening candidate modeling samples with similarity indexes larger than preset similarity indexes with the target modeling sample from the candidate modeling sample set based on the similarity indexes, so as to obtain a plurality of modeling samples.
In an optional embodiment of the present application, the calculating module 1103 is specifically configured to: the target power load prediction model is
Figure BDA0002823632720000171
In the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000172
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t)-1),…s(t-lm) Is historical load data; lpTo predict the step size, if l p1, i.e. single step prediction,/p>1 is the multi-step prediction, and the prediction method,
Figure BDA0002823632720000173
the output of the model is predicted for the target electrical load.
In an optional embodiment of the present application, the calculating module 1103 is specifically configured to: based on the formula
Figure BDA0002823632720000181
Calculating parameters of a target power load prediction model; in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000182
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) For historical load data, k is the number of modeling samples in the set of modeling samples, wi＝1-D(φi,φt) The weight of the modeled sample, D (phi), determined for the similarity indexi,φt) Modeling a sample phi for the targettAnd a modeling sample phiiSimilarity index between, yiFor the modeling sample
Figure BDA0002823632720000183
The corresponding power load value.
In an optional embodiment of the present application, the predicting module 1104 is specifically configured to: acquiring a previous momentm(ii) historical load data; constructing a target power load prediction model based on parameters of the target power load prediction model; the current time is the previous onemThe historical load data of each time series is input into a target power load prediction model, and power load data after the predicted current time is obtained.
For specific limitations of the power load prediction apparatus, reference may be made to the above limitations of the power load prediction method, which are not described herein again. Each module in the above power load prediction apparatus may be wholly or partially implemented by software, hardware, or a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program: acquiring m pieces of historical load data in a preset time period before the current moment, wherein m is an integer greater than 1; obtaining a modeling sample set from m pieces of historical load data, wherein the modeling sample set comprises a plurality of modeling samples, and each modeling sample comprises lmThe historical load data in each modeling sample has the relevance larger than a preset relevance threshold value lmIs an integer greater than 1; calculating parameters of a target power load prediction model based on the modeling sample set; and predicting the power load data after the current moment based on the parameters of the target power load prediction model and the target power load prediction model.
In an alternative embodiment of the application, the processor when executing the computer program further performs the step of calculating a time series length l of the target modeled samplemLength of time series lmThe maximum number of the time-continuous historical load data with correlation in the m pieces of historical load data is represented; based on the length of the time series lmAnd acquiring a plurality of modeling samples from the m pieces of historical load data to construct a modeling sample set.
In an alternative embodiment of the application, the processor when executing the computer program further performs the steps of: based on the partial autocorrelation coefficients, sequentially calculating the correlation degree between the current load data and the historical load data which is continuous in time before the current time from the load data corresponding to the current time in the m pieces of historical load data; determining a time sequence of the target modeling samples when the correlation is less than a preset correlation valueLength of column lm。
In an alternative embodiment of the application, the processor when executing the computer program further performs the steps of: determining a candidate modeling sample set based on the m historical load data, wherein the candidate modeling sample set comprises a plurality of candidate modeling samples; respectively calculating similarity indexes of the target modeling sample and a plurality of candidate modeling samples in the candidate modeling sample set; and screening candidate modeling samples with similarity indexes larger than preset similarity indexes with the target modeling sample from the candidate modeling sample set based on the similarity indexes, so as to obtain a plurality of modeling samples.
In an alternative embodiment of the application, the processor when executing the computer program further performs the steps of:
the target power load prediction model is as follows:
Figure BDA0002823632720000191
in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000192
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) Is historical load data; lpTo predict the step size, if l p1, i.e. single step prediction,/p>1 is the multi-step prediction, and the prediction method,
Figure BDA0002823632720000193
the output of the model is predicted for the target electrical load.
In an alternative embodiment of the application, the processor when executing the computer program further performs the steps of:
based on the formula
Figure BDA0002823632720000194
Calculating parameters of a target power load prediction model; in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000195
is a target electricityParameter of the force load prediction model, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) For historical load data, k is the number of modeling samples in the set of modeling samples, wi＝1-D(φi,φt) The weight of the modeled sample, D (phi), determined for the similarity indexi,φt) Modeling a sample phi for the targettAnd a modeling sample phiiSimilarity index between, yiFor the modeling sample
Figure BDA0002823632720000196
The corresponding power load value.
In an alternative embodiment of the application, the processor when executing the computer program further performs the steps of: acquiring a previous momentm(ii) historical load data; constructing a target power load prediction model based on parameters of the target power load prediction model; the current time is the previous onemThe historical load data is input to a target power load prediction model, and predicted power load data after the current time is obtained.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring m pieces of historical load data in a preset time period before the current moment, wherein m is an integer greater than 1; obtaining a modeling sample set from m pieces of historical load data, wherein the modeling sample set comprises a plurality of modeling samples, and each modeling sample comprises lmThe historical load data in each modeling sample has the relevance larger than a preset relevance threshold value lmIs an integer greater than 1; calculating parameters of a target power load prediction model based on the modeling sample set; and predicting the power load data after the current moment based on the parameters of the target power load prediction model and the target power load prediction model.
In an alternative embodiment of the application, the computer program is further arranged to be executed by a processorThe method comprises the following steps of calculating the time sequence length l of the target modeling samplemLength of time series lmThe maximum number of the time-continuous historical load data with correlation in the m pieces of historical load data is represented; based on the length of the time series lmAnd acquiring a plurality of modeling samples from the m pieces of historical load data to construct a modeling sample set.
In an alternative embodiment of the application, the computer program when executed by the processor further performs the steps of:
based on the partial autocorrelation coefficients, sequentially calculating the correlation degree between the current load data and the historical load data which is continuous in time before the current time from the load data corresponding to the current time in the m pieces of historical load data; when the correlation degree is smaller than a preset correlation degree value, determining the time sequence length l of the target modeling samplem。
In an alternative embodiment of the application, the computer program when executed by the processor further performs the steps of:
determining a candidate modeling sample set based on the m historical load data, wherein the candidate modeling sample set comprises a plurality of candidate modeling samples; respectively calculating similarity indexes of the target modeling sample and a plurality of candidate modeling samples in the candidate modeling sample set; and screening candidate modeling samples with similarity indexes larger than preset similarity indexes with the target modeling sample from the candidate modeling sample set based on the similarity indexes, so as to obtain a plurality of modeling samples.
In an alternative embodiment of the application, the computer program when executed by the processor further performs the steps of:
the target power load prediction model is as follows:
Figure BDA0002823632720000211
in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000212
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) Is historical load data; lpTo predict the step size, if l p1, i.e. single step prediction,/p>1 is the multi-step prediction, and the prediction method,
Figure BDA0002823632720000213
the output of the model is predicted for the target electrical load.
In an alternative embodiment of the application, the computer program when executed by the processor further performs the steps of:
based on the formula
Figure BDA0002823632720000214
Calculating parameters of a target power load prediction model; in the formula (I), the compound is shown in the specification,
Figure BDA0002823632720000215
parameters of the prediction model for the target power load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of a target power load prediction model, s (t), s (t-1), … s (t-l)m) For historical load data, k is the number of modeling samples in the set of modeling samples, wi＝1-D(φi,φt) The weight of the modeled sample, D (phi), determined for the similarity indexi,φt) Modeling a sample phi for the targettAnd a modeling sample phiiSimilarity index between, yiFor the modeling sample
Figure BDA0002823632720000216
The corresponding power load value.
In an alternative embodiment of the application, the computer program when executed by the processor further performs the steps of: acquiring a previous momentmHistorical load data of time succession; constructing a target power load prediction model based on parameters of the target power load prediction model; the current time is the previous onemThe historical load data is input to a target power load prediction model, and predicted power load data after the current time is obtained.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (10)

1. A method of predicting a power load, the method comprising:
acquiring m pieces of historical load data in a preset time period before the current moment, wherein m is an integer greater than 1;
obtaining a set of modeling samples from the m historical load data, theThe modeling sample set comprises a plurality of modeling samples, and each modeling sample comprises lmHistorical load data with continuous time, the correlation of the historical load data in each modeling sample is larger than a preset correlation threshold value,/mIs an integer greater than 1;
calculating parameters of a target power load prediction model based on the modeling sample set;
predicting the power load data after the current time based on the parameters of the target power load prediction model and the target power load prediction model.
2. The method of claim 1, wherein obtaining a set of modeling samples from the m historical load data comprises:
calculating the time series length l of the target modeling samplemLength of said time series lmThe maximum number of the time-continuous historical load data with correlation in the m pieces of historical load data is represented;
based on the time series length lmAnd acquiring a plurality of modeling samples from the m pieces of historical load data to construct the modeling sample set.
3. The method of claim 2, wherein the calculating the time series length/of the target sample setmThe method comprises the following steps:
based on the partial autocorrelation coefficient, sequentially calculating the correlation degree between the current load data and the historical load data which is continuous in time before the current time from the load data corresponding to the current time;
when the correlation degree is smaller than a preset correlation degree value, determining the time sequence length l of the target modeling samplem。
4. The method of claim 2, wherein the time sequence is based on the length lmObtaining a plurality of said modeling samples from said m historical load dataThe method comprises the following steps:
determining a set of candidate modeling samples based on the m historical load data, wherein the set of candidate modeling samples comprises a plurality of candidate modeling samples;
calculating similarity indicators of the target modeling sample and the plurality of candidate modeling samples in the candidate modeling sample set respectively;
and screening candidate modeling samples with similarity indexes larger than preset similarity indexes with the target modeling sample from the candidate modeling sample set based on the similarity indexes, so as to obtain a plurality of modeling samples.
5. The method of claim 1, wherein the target electrical load prediction model is:
Figure FDA0002823632710000021
in the formula (I), the compound is shown in the specification,
Figure FDA0002823632710000022
predicting a parameter of a model for the target electrical load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of the target power load prediction model, s (t), s (t-1), … s (t-l)m) Is historical load data; lpTo predict the step size, if lp1, i.e. single step prediction,/p>1 is the multi-step prediction, and the prediction method,
Figure FDA0002823632710000023
predicting an output of the model for the target electrical load.
6. The method of claim 1, wherein said calculating parameters of a target power load prediction model based on said modeled sample set comprises:
based on the formula
Figure FDA0002823632710000024
Calculating parameters of the target power load prediction model;
in the formula (I), the compound is shown in the specification,
Figure FDA0002823632710000025
predicting a parameter of a model for the target electrical load, phit＝(s(t),s(t-1),…s(t-lm) Is a variable of the target power load prediction model, s (t), s (t-1), … s (t-l)m) For historical load data, k is the number of modeling samples in the set of modeling samples, wi＝1-D(φi,φt) The weight of the modeled sample, D (phi), determined for the similarity indexi,φt) Modeling a sample phi for the targettAnd a modeling sample phiiSimilarity index between, yiFor the modeling sample
Figure FDA0002823632710000026
The corresponding power load value.
7. The method of claim 1, wherein predicting power load data after the current time based on the parameters of the target power load prediction model and the target power load prediction model comprises:
acquiring a previous momentm(ii) historical load data;
constructing the target power load prediction model based on the parameters of the target power load prediction model;
l is prior to the current timemAnd inputting the historical load data with continuous time into the target power load prediction model to obtain the predicted power load data after the current time.
8. An electrical load prediction apparatus, the apparatus comprising:
the first acquisition module is used for acquiring m pieces of historical load data in a preset time period before the current moment, wherein m is an integer greater than 1;
a second obtaining module, configured to obtain a modeling sample set from the m pieces of historical load data, where the modeling sample set includes a plurality of modeling samples, and each modeling sample includes lmHistorical load data, the correlation of the historical load data in each modeling sample is greater than a preset correlation threshold value,/mIs an integer greater than 1;
a calculation module for calculating parameters of a target power load prediction model based on the set of modeling samples;
and the prediction module is used for predicting the power load data after the current moment based on the parameters of the target power load prediction model and the target power load prediction model.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
CN202011423514.XA 2020-12-08 2020-12-08 Power load prediction method, power load prediction device, computer equipment and storage medium Pending CN112465250A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011423514.XA CN112465250A (en) 2020-12-08 2020-12-08 Power load prediction method, power load prediction device, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011423514.XA CN112465250A (en) 2020-12-08 2020-12-08 Power load prediction method, power load prediction device, computer equipment and storage medium

Publications (1)

Publication Number Publication Date
CN112465250A true CN112465250A (en) 2021-03-09

Family

ID=74800988

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011423514.XA Pending CN112465250A (en) 2020-12-08 2020-12-08 Power load prediction method, power load prediction device, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN112465250A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113672666A (en) * 2021-08-23 2021-11-19 成都佳华物链云科技有限公司 Machine load prediction method and device, electronic equipment and readable storage medium
CN114048362A (en) * 2022-01-11 2022-02-15 国网电子商务有限公司 Block chain-based power data anomaly detection method, device and system

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107506868A (en) * 2017-08-31 2017-12-22 广东工业大学 A kind of method and device of temporary electricity load prediction
CN110570023A (en) * 2019-08-16 2019-12-13 国网天津市电力公司 short-term commercial power load prediction method based on SARIMA-GRNN-SVM
CN110648026A (en) * 2019-09-27 2020-01-03 京东方科技集团股份有限公司 Prediction model construction method, prediction method, device, equipment and medium
CN111160625A (en) * 2019-12-10 2020-05-15 中铁电气化局集团有限公司 Power load prediction method, power load prediction device, computer equipment and storage medium
CN111340273A (en) * 2020-02-17 2020-06-26 南京邮电大学 Short-term load prediction method for power system based on GEP parameter optimization XGboost

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107506868A (en) * 2017-08-31 2017-12-22 广东工业大学 A kind of method and device of temporary electricity load prediction
CN110570023A (en) * 2019-08-16 2019-12-13 国网天津市电力公司 short-term commercial power load prediction method based on SARIMA-GRNN-SVM
CN110648026A (en) * 2019-09-27 2020-01-03 京东方科技集团股份有限公司 Prediction model construction method, prediction method, device, equipment and medium
CN111160625A (en) * 2019-12-10 2020-05-15 中铁电气化局集团有限公司 Power load prediction method, power load prediction device, computer equipment and storage medium
CN111340273A (en) * 2020-02-17 2020-06-26 南京邮电大学 Short-term load prediction method for power system based on GEP parameter optimization XGboost

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
马立新 等: "SOM特征提取与ELM在电力负荷预测中的应用", 电力科学与工程, vol. 31, no. 5, pages 1 - 5 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113672666A (en) * 2021-08-23 2021-11-19 成都佳华物链云科技有限公司 Machine load prediction method and device, electronic equipment and readable storage medium
CN114048362A (en) * 2022-01-11 2022-02-15 国网电子商务有限公司 Block chain-based power data anomaly detection method, device and system

Similar Documents

Publication Publication Date Title
CN112699913B (en) Method and device for diagnosing abnormal relationship of household transformer in transformer area
Tong et al. A hybrid algorithm for reliability analysis combining Kriging and subset simulation importance sampling
CN108959778B (en) Method for predicting residual life of aircraft engine based on consistency of degradation modes
CN107506868B (en) Method and device for predicting short-time power load
US20180046902A1 (en) Enhanced restricted boltzmann machine with prognosibility regularization for prognostics and health assessment
CN114285728B (en) Predictive model training method, traffic prediction device and storage medium
JPWO2017094267A1 (en) Anomaly detection system, anomaly detection method, anomaly detection program, and learned model generation method
WO2015196133A2 (en) Energy infrastructure sensor data rectification using regression models
JP2009294969A (en) Demand forecast method and demand forecast device
CN114297036B (en) Data processing method, device, electronic equipment and readable storage medium
Goodman et al. Prognostics and health management: A practical approach to improving system reliability using condition-based data
TWI738597B (en) Analog battery construction method and analog battery construction device
CN112465250A (en) Power load prediction method, power load prediction device, computer equipment and storage medium
CN107944612A (en) A kind of busbar net load Forecasting Methodology based on ARIMA and phase space reconfiguration SVR
JP6086875B2 (en) Power generation amount prediction device and power generation amount prediction method
CN115392301A (en) Converter transformer state identification method, converter transformer state identification device, converter transformer state identification equipment, converter transformer state identification medium and program product
CN108694472B (en) Prediction error extreme value analysis method, device, computer equipment and storage medium
CN114970665A (en) Model training method, electrolytic capacitor residual life prediction method and system
CN113110961B (en) Equipment abnormality detection method and device, computer equipment and readable storage medium
CN117076931B (en) Time sequence data prediction method and system based on conditional diffusion model
CN114118570A (en) Service data prediction method and device, electronic equipment and storage medium
CN115935244B (en) Single-phase rectifier fault diagnosis method based on data driving
CN114692529A (en) CFD high-dimensional response uncertainty quantification method and device, and computer equipment
CN115908051A (en) Method for determining energy storage capacity of power system
Bosma et al. Estimating solar and wind power production using computer vision deep learning techniques on weather maps

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