CN110942195A - Power load prediction method and device - Google Patents
Power load prediction method and device Download PDFInfo
- Publication number
- CN110942195A CN110942195A CN201911178565.8A CN201911178565A CN110942195A CN 110942195 A CN110942195 A CN 110942195A CN 201911178565 A CN201911178565 A CN 201911178565A CN 110942195 A CN110942195 A CN 110942195A
- Authority
- CN
- China
- Prior art keywords
- data
- type
- power
- prediction
- external factors
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 230000001364 causal effect Effects 0.000 claims abstract description 49
- 238000012795 verification Methods 0.000 claims abstract description 27
- 230000005611 electricity Effects 0.000 claims description 45
- 238000012549 training Methods 0.000 claims description 32
- 239000011159 matrix material Substances 0.000 claims description 12
- 230000002195 synergetic effect Effects 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 4
- 238000009826 distribution Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims 2
- 238000010586 diagram Methods 0.000 description 10
- 238000001556 precipitation Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000000556 factor analysis Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 235000019580 granularity Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
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
- 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
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The application discloses a method and a device for predicting electrical load. According to the method, aiming at each power consumption type, power consumption data and external factor data of the type in a preset time period are obtained, causal verification is conducted on the power consumption data and the external factor data, the fact-and-effect relation between N external factors and the power consumption type is determined, M external factors are selected from the N external factors in a wrapping type feature selection mode, a prediction model is trained according to the M external factor data and the power consumption data of the power consumption type, the power consumption load of the type is predicted according to the prediction model, and a prediction result is output; the total power load can be predicted according to the prediction result of each power consumption type. By the method, the accuracy of the electric load is improved.
Description
Technical Field
The present application relates to the field of power technologies, and in particular, to a method and an apparatus for predicting a power consumption load.
Background
The load prediction is an important basis of power grid dispatching operation, and accurate load prediction plays an important role in improving the arrangement of a power grid operation mode and improving the operation benefit of a power system. Particularly, along with the continuous deepening of the reform of the power market in China in recent years, the demand of the decision analysis of market members on high-precision load prediction is more urgent.
In the traditional mode, a correlation factor analysis method is a main method for load prediction research. The method mainly realizes load prediction by analyzing the correlation between the data sequence of the relevant factors such as precipitation, sunshine and the like and the power load and based on the prediction data of the relevant factors. The analysis method of the correlation between the relevant factors and the electric load evolves from the primary/secondary function fitting step by step to the neural network and other methods. However, in the face of the change of the electricity load characteristics in the new and normal economic development and the continuous improvement of the load prediction accuracy requirement in China under the reform of the electricity market, the traditional correlation factor analysis method faces the following challenges:
on the other hand, economic development of China enters a new normal state, the economic growth speed, the economic development structure, the economic driving mode and the like are obviously changed, so that the internal rule of the power load is changed, and the traditional power load prediction method cannot meet the current requirement on the power load prediction accuracy.
Disclosure of Invention
The embodiment of the application provides an electric load prediction method and equipment, which are used for realizing accurate prediction of an electric load.
In a first aspect, a method for predicting an electrical load provided in an embodiment of the present application includes:
acquiring power load data of a target power utilization type in a preset time period and data of various external factors in the preset time period;
carrying out causal verification on the power load data of the target power utilization type and the multiple external factor data, and determining that N external factors in the multiple external factors have causal relation with the power load of the target power utilization type;
selecting M external factors from the N external factors in a wrapping type feature selection mode, and training a prediction model according to the data of the M external factors and the power load data of the target power utilization type;
and predicting the power load of the target power utilization type according to the prediction model, and outputting a prediction result.
In a possible implementation manner, selecting M external factors from the N external factors in a wrapping-type feature selection manner, and training a prediction model according to data of the M external factors and power load data of the target power consumption type, includes:
sorting the N external factors according to the checking result, wherein the higher the sorting is, the higher the correlation degree between the external factors and the power load of the target power consumption type is;
selecting the previous X external factors, training a first prediction model according to the X external factor data and the power load data of the target power utilization type, and estimating the accuracy of the first prediction model, wherein X is a preset constant;
training a second prediction model according to the previous M external factor data and the power load data of the target power utilization type, and estimating the accuracy of the second prediction model, wherein M is X + 1; if the value obtained by subtracting the accuracy of the first prediction model from the accuracy of the second prediction model meets the preset condition, taking the second prediction model as the first prediction model, making M equal to M +1, and repeatedly executing the step of training the second model until the value obtained by subtracting the accuracy of the first prediction model from the accuracy of the second prediction model does not meet the preset condition, or until M +1 is greater than N;
the first prediction model is used for predicting the target electricity utilization type.
In one possible implementation, the causally verifying the power load data of the target power usage type and the data of the plurality of external factors includes:
performing coordinated calibration according to the power load data of the target power utilization type and the data of the multiple external factors;
and carrying out causal verification on the data of the external factors meeting the synergistic relationship and the power load data of the target power utilization type.
In one possible implementation, the parity check is a johnson parity check; and/or, the causal check is a granger causal check.
In one possible implementation, the performing johnson coordination check according to the power load data of the target power usage type and each external factor data of the external factors includes:
for each external factor, the following steps are performed:
based on the electrical load data and the external factor data, a correction model is constructed according to the following formula:
wherein, Delta YtIs a difference term of the target sequence, Δ Yt-iIs the i-order lag difference term, Γ, of the target sequenceiCoefficient matrix of difference values of i-order lag terms, ΠpIs a matrix of lag term coefficients of order p, UtIs a residual error;
according to IIpDetermines whether the external factor and the power load of the target power usage type satisfy a synergistic relationship.
In one possible implementation, the performing the granger causal check on the data of the external factors satisfying the synergistic relationship and the power load data of the target power usage type includes:
for each external factor satisfying the co-integration relationship, the following steps are performed:
constructing a time series vector autoregressive model based on the target type of the power load data and the external factor data;
time series Y to be evaluated1tAnd the external factor time series Y2tComputing residual sum of squares by least squares estimation
Time series Y to be evaluated1tComputing the sum of squares of the residuals by least squares estimation
Statistic F is calculated according to the following formula:
wherein r is zero constraint number, T is sample capacity, and p is lag period number of the vector autoregressive model;
based on x2(r) calculating rF and χ by distribution model2(r) critical value if rF.gtoreq.X2(r), the external factor has a causal relationship with the target electricity usage type.
In a second aspect, an embodiment of the present application provides an electrical load prediction method, including:
for each power consumption type, performing power consumption load prediction according to the power consumption load prediction method of the first aspect;
and predicting the total power load according to the power load prediction result of each power consumption type, and outputting the prediction result.
In a third aspect, an embodiment of the present application provides an electrical load prediction apparatus, including:
the input module is used for acquiring power load data of a target power utilization type in a preset time period and data of various external factors in the preset time period;
the prediction module is used for carrying out causal verification on the power load data of the target power utilization type and the multiple external factor data and determining that N external factors in the multiple external factors have causal relationship with the power load of the target power utilization type; selecting M external factors from the N external factors in a wrapping type feature selection mode, and training a prediction model according to the data of the M external factors and the power load data of the target power utilization type; predicting the power load of the target power utilization type according to the prediction model;
and the output module is used for outputting the prediction result.
In a fourth aspect, an embodiment of the present application provides an electrical load prediction apparatus, including:
the input module is used for acquiring power load data of each power utilization type in a preset time period and various external factor data in the preset time period;
a prediction module, configured to perform power consumption load prediction according to the power consumption load prediction method of the first aspect, for each power consumption type; predicting the total power load according to the power load prediction result of each power consumption type;
and the output module is used for outputting the prediction result.
In a fifth aspect, the present embodiments provide a computer-readable storage medium storing computer instructions, which, when executed on a computer, cause the computer to perform the functions of the method according to any one of the first and second aspects.
In the traditional power load prediction process, the difference of different power types is not considered, the influence difference of the same external factor on different power types is not considered, and the total power load is predicted only according to various external factors; however, the correlation factors of different electricity utilization types are different, and the different correlation factors may have the characteristics of coverage, cancellation and the like, so that the accuracy of prediction is high instead of selecting multiple factors with high correlation degree with the electricity utilization load. After the embodiment of the application is applied, the power utilization types are respectively predicted, only the external factors related to the power utilization loads of the types are considered in the prediction process, and the external factors are selected in a wrapping type feature selection mode, so that the problems of repeated calculation, accuracy reduction and the like caused by the characteristics of coverage, offset and the like among the different factors are solved, and a more accurate prediction result of the power utilization loads of the target power utilization types is obtained.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic flow chart of a power load prediction method according to an embodiment of the present disclosure;
fig. 2 is a second schematic flow chart of a power load prediction method according to an embodiment of the present application;
fig. 3 is a third schematic flow chart of a power load prediction method according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of an electrical load prediction apparatus according to an embodiment of the present application;
fig. 5 is a second schematic structural diagram of an electrical load prediction apparatus according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the technical solutions of the present application will be described in detail below. It is to be understood that the embodiments described are only a few embodiments of the present application and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the examples given herein without making any creative effort, shall fall within the protection scope of the present application.
Along with the adjustment of economic development structure of China, the electrical loads of different industries are subjected to structure transformation. In the traditional mode, the power load is generally regarded as a whole, the power utilization characteristics of different industries are not considered, and the influence of different combination modes of related factors on the prediction precision is not considered, so that the prediction result and the prediction accuracy are naturally difficult to meet the actual requirements.
The embodiment of the application provides an electricity load prediction method, which is used for realizing electricity load prediction with higher precision and is beneficial to optimizing electricity capacity.
Referring to fig. 1, a flow chart of a method for predicting an electrical load according to an embodiment of the present disclosure is schematically shown, and as shown in the drawing, the method may include the following steps:
In order to obtain a relatively accurate prediction result, the embodiment of the application respectively predicts different power utilization types. For example, the electric load may be classified into an agricultural electric load, an industrial electric load, a residential electric load, and the like, and when prediction is performed for each type of electric load, the electric load data of the corresponding type may be acquired. It should be understood that the foregoing division of the power utilization type is only an example, and the type division with different angles and different granularities may be performed according to actual requirements, which is not limited in the embodiment of the present application.
The external factors may include various factors that may be associated with the power load, such as temperature, solar radiation intensity, precipitation, wind speed, power price, relevant policy, and the like. For example, for residential electricity consumption, when the temperature is too high, the utilization rate of an air conditioner and a fan is increased, which may cause the increase of the electrical load of residents; when there is large precipitation, the agricultural work activity is reduced, possibly resulting in a reduction in the agricultural electrical load.
The acquired electric load data and the external factor data are corresponding data in time. For example, data such as electricity load, air temperature, precipitation, solar radiation intensity, and wind speed at 8 to 18 points per day on yesterday, the previous day, and the most previous day are acquired.
In a specific embodiment, the acquired power load data and each external factor are time series, and the data at the same position in different time series respectively represent the power load or the external factor data at the same time. For example, the acquired power load data is [ P1, P2, P3, … ], the acquired data of the external factor-air temperature is [ T1, T2, T3, … ], the acquired data of the external factor-solar intensity is [ S1, S2, S3, … ], the acquired data of the external factor-precipitation amount is [ M1, M2, M3, … ], wherein P1, T1, S1, M1 respectively indicate the power load, air temperature, solar intensity, precipitation amount in a T1 period, P2, T2, S2, M2 respectively indicate the power load, air temperature, solar intensity, precipitation amount in a T2 period, and P3, T3, S3, M3 respectively indicate the power load, solar intensity, precipitation amount in a T3 period.
102, carrying out causal verification on the power load data of the target power utilization type and multiple external factors, and determining that the N external factors in the multiple external factors have a causal relationship with the power load of the target power utilization type.
The external factor involved in the above step 101 may be data of a factor considered to be possibly related to the electric load, for example, a factor considered to be related to the electric load in the conventional mode, or a factor considered to be possibly related to the electric load as the electric structure is changed, or the like. However, all the external factors involved in step 101 are not necessarily related to the electric load of the target electricity usage type, and therefore, whether each external factor has a causal relationship with the electric load of the target electricity usage type can be determined by a causal check.
In a possible implementation mode, the coordinated calibration can be carried out according to the power load data of the target power utilization type and the data of various external factors; and further performing causal verification according to the external factor data and the target power consumption type power consumption load data aiming at the external factors meeting the cooperative relationship, so as to determine which external factors and the target power consumption load have causal relationship in the plurality of external factors involved in the step 101.
Optionally, when performing the coordination check, a johanson coordination check method may be adopted. In the causal verification, a glange causal verification method can be adopted.
Specifically, when the johanson coordination check is adopted, the following steps may be respectively performed for each external factor to be checked:
1) and constructing a correction model according to the formula (1) based on the power load data of the target power utilization type and the external factor data to be verified.
Wherein, YtFor a target sequence, i.e. using the electrical load time series, Δ YtIs a difference term of the target sequence, Δ Yt-iIs the i-order lag difference term, Γ, of the target sequenceiFor the i-th order lag term difference coefficient matrix, Π, constructed according to external factorspIs a matrix of lag term coefficients of order p, UtIs the residual error.
2) According to IIpDetermines whether the external factor and the power load of the target power usage type satisfy a synergistic relationship.
The co-integration relation means that stable equilibrium relation exists among the non-stationary time sequences, and the purpose of the co-integration check is to judge whether stable equilibrium relation exists among a group of non-stationary time sequences. Because the electrical load data and the external factor data are often non-stationary time sequences, and pseudo regression is likely to occur in the non-stationary sequences, whether the causal relationship described by the regression equation is pseudo regression or not can be judged through the co-integration check, namely whether a stable relationship exists between the electrical load and the external factor or not is judged.
The above coefficient matrix pipThe rank of (b) can be used to characterize whether the corresponding sequence has a co-integration relationship with the target sequence, specifically, an r-order coefficient matrix pipCorresponding sequence Yt-1、Yt-2、…、Yt-rWith the target sequence YtHaving a co-integration relationship, i.e. sequence Yt-1、Yt-2、…、Yt-rThe corresponding external factors have a synergistic relationship with the target type of electrical load. In particular, if the coefficient matrix ΠpThe full rank indicates that the adopted external factors have a synergistic relationship with the target type of the power load; if coefficient matrix pipThe rank of (1) is 0, which indicates that no coordination relationship exists between the adopted external factors and the target type of the power load.
And after determining whether the external factors and the target type power load have the co-integration relation, performing the Glan's causal check on the data of the external factors and the data of the target type power load meeting the co-integration relation.
Specifically, when the grand cause-and-effect verification is adopted, the following steps may be respectively performed for each external factor to be verified:
1) and constructing n time series vector autoregressive models based on the target type electric load data and the external factor data to be verified. Where n represents the number of external factors to be checked, i.e., the number of external factors satisfying the co-integration relationship.
2) To electrical load sequence Y1tAnd external factor time series Y2tComputing residual sum of squares by least squares estimation
3) Time series Y to be evaluated1tComputing the sum of squares of the residuals by least squares estimation
4) Calculate statistic F according to equation (2):
wherein r is zero constraint number; t is the sample volume; p is the lag phase number of the vector autoregressive model, and can be obtained by calculation through methods such as LR (least squares) verification and Akaike Information Criterion (AIC).
5) Based on x2(r) calculating rF and χ by distribution model2(r) critical value if rF.gtoreq.X2(r), the external factor has a causal relationship with the target electricity usage type.
After the Glankey cause-and-effect verification, which external factors have cause-and-effect relationship with the target power utilization type power utilization load can be determined. The external factors causal to the target electricity load may be used to predict the target electricity load, while the external factors without causal relationship are not considered in the prediction.
And 103, selecting M external factors from the N external factors in a wrapping type feature selection mode, and training a prediction model according to data of the M external factors and power load data of the target power utilization type.
Assuming that it is determined in step 102 that there is a causal relationship between the N external factors and the target power consumption type power load, when predicting the target power consumption type power load, the N external factors are not necessarily considered, M external factors may be selected from the N external factors, and the target power consumption type power load may be predicted according to the M external factors. For example, the prediction may be performed only by using an external factor having a high correlation with the power load of the target power usage type; for example, the target electricity usage type electricity load may be predicted using a larger variety of external factors, and the prediction accuracy may not be significantly improved.
In one possible implementation, the M external factors may be selected and the predictive model may be trained as follows:
1) the N external factors having a causal relationship with the target electricity consumption type electricity load are ranked, and the higher the ranking is, the higher the correlation between the external factors and the target electricity consumption type electricity load is.
As mentioned above, the causal verification of multiple external factors can be performed by Johnson synergy verification and Greenjek causal verification, and after the Greenjek causal verification, the rF and the χ can be determined according to the rF and the χ2(r) the difference determines the degree of correlation between the external factor and the power load of the target power consumption type, i.e., rF and χ2The greater the difference (r) is, the higher the degree of correlation between the external factor and the power load of the target power consumption type is, and the higher the correlation can be made from rF and χ2The difference of (r) ranks the N external factors.
If other verification modes are adopted, the N external factors can be sequenced according to the characteristics of the other verification modes; alternatively, the sorting may be performed based on prior experience.
2) Selecting X external factors ranked in the front, training a first prediction model according to the data of the X external factors and the data of the target power utilization type power utilization load, and estimating the accuracy of the first prediction model.
Wherein X is a preset constant. For example, the top-ranked external factors in selection 3 may be set to be used for training the first prediction model.
When the first prediction model is trained, prediction can be performed according to artificial intelligence algorithms such as a neural network and a support vector machine, and the algorithm for training the prediction model is not limited in the embodiment of the application.
After the first prediction model is trained, the accuracy of the first prediction model can be estimated according to the verification sample data. Specifically, the data acquired in step 101 may include training sample data and verification sample data; or the obtained data does not distinguish the training sample from the verification sample, and the obtained data can be divided according to a preset proportion after the data is obtained, so that the training sample data and the verification sample data are obtained.
In one possible implementation, the accuracy of the first predictive model may be estimated according to equation (3).
Wherein, ymapeIs an average absolute percent error prediction index; n is the number of verification sample data, Xact(i) And Xpred(i) The actual electric load and the predicted electric load at the ith prediction point are respectively represented.
3) And training a second prediction model according to the data of the M external factors and the power load data of the target power utilization type, wherein M is X +1, and estimating the accuracy of the second prediction model.
The way of training the second prediction model is consistent with the way of training the first model, and the way of estimating the accuracy of the second prediction model is consistent with the way of estimating the accuracy of the first prediction model, and the difference is only data added with one external factor.
4) If the value obtained by subtracting the accuracy of the first prediction model from the accuracy of the second prediction model meets the preset condition, taking the second prediction model as the first prediction model, and enabling M to be M + 1; and repeatedly executing the step 3) and the step 4) until the value obtained by subtracting the accuracy of the first prediction model from the accuracy of the second prediction model does not meet the preset condition, or until M +1> N.
For example, a threshold may be preset, and if the difference between the accuracy of the second prediction model and the accuracy of the first prediction model is greater than or equal to the preset threshold, it may be considered that the accuracy of predicting the power load of the target power consumption type is effectively improved after an external factor is added. If the difference value is smaller than the preset threshold value, the accuracy of the current first prediction model can be considered to be converged, namely, the accuracy is not obviously improved by adding an external factor; since each additional external factor will increase the calculation amount or make the calculation process more complicated, it is not necessary to increase the types of external factors when the accuracy has converged.
When the repeated execution of the steps 3) and 4) is stopped, the first prediction model at the moment is a prediction model for predicting the target electricity utilization type.
And step 104, predicting the power load of the target power utilization type according to the prediction model, and outputting a prediction result.
After the prediction model is determined, the power load of the target power utilization type can be predicted according to the data of the M external factors used for training the prediction model, so that a relatively accurate prediction result is obtained.
When the prediction result is output, the prediction result may be displayed to a user through a display device, or the prediction result may also be output to other devices, so that the other devices perform subsequent operations according to the prediction result, for example, summarize various prediction results, and perform optimization adjustment on the electric power production according to the prediction result, or the like.
In the traditional power load prediction process, the difference of different power types is not considered, the influence difference of the same external factor on different power types is not considered, and the total power load is predicted only according to various external factors; however, the correlation factors of different electricity utilization types are different, and the different correlation factors may have the characteristics of coverage, cancellation and the like, so that the accuracy of prediction is high instead of selecting multiple factors with high correlation degree with the electricity utilization load. In the embodiment of the application, the power consumption types are respectively predicted, only the external factors related to the power consumption loads of the types are considered in the prediction process, and the external factors are selected in a wrapping type feature selection mode, so that the problems of repeated calculation, accuracy reduction and the like caused by characteristics such as coverage and offset among the different factors are solved, and a more accurate prediction result of the power consumption loads of the target power consumption types is obtained.
Based on the same technical concept, an embodiment of the present application further provides an electrical load prediction method, as shown in fig. 2, the method may include the following steps:
Specifically, the power load data acquired from the outside may be all power load data, and then the power load data is classified, so that power load data of each power consumption type is acquired; alternatively, the data acquired from the outside may be power load data that has been divided according to the power usage type.
The power load prediction for each power consumption type in step 202 is similar to the power load prediction method in the foregoing embodiment, and is not described herein again.
And 203, predicting the total power load according to the power load prediction result of each power consumption type, and outputting the prediction result.
When the total power load is predicted, the total power load can be determined in an accumulation mode. For example, the total electrical load may be determined by equation (4).
Wherein,
represents a predicted value of the total electricity load in the t-th period on the d-th day, ECN represents the number of electricity types,
and the predicted value of the power load of the ith power utilization type in the t period on the day d is shown.
Of course, the above method for directly accumulating the prediction results of the electrical loads of each electrical type to determine the total electrical load is only a specific embodiment, and the total electrical load may also be predicted in other manners, for example, the prediction results of the electrical types may be set with corresponding weights according to the accuracy of the prediction model of each electrical type.
In an embodiment, the flow of predicting the total power load may be as shown in fig. 3, and specifically includes:
and 301, acquiring data of the electric load and data of various external factors.
And step 302, dividing the power consumption load data according to the power consumption types to obtain the power consumption load data of each power consumption type.
Next, the prediction is performed for each electricity consumption type, that is, steps 303 to 3011 are performed, and the prediction of the electricity load of electricity consumption type 1 will be described as an example.
Step 303(1), performing Johanson coordination check on the electricity load data of the electricity type 1 and various external data.
And step 304(1) of performing a granger causal check on the data of the external factors satisfying the co-integration relation and the power load data of the power type 1 to obtain an external factor set (hereinafter referred to as a set) having a causal relation with the power load of the power type 1.
And step 305(1), sorting the external factors in the set according to the relevance, wherein the higher the relevance is, the higher the ranking is.
And step 306(1), training a first prediction model according to the data of the external factors ranked in the first three digits and the power load data of the power type 1, and estimating the accuracy of the first prediction model.
And 307(1) training a second prediction model according to the data with the sequence added with an external factor, and estimating the accuracy of the second prediction model.
And 308, (1) judging whether the accuracy rate is converged, namely, judging whether the difference value between the accuracy rate of the second prediction model and the accuracy rate of the first prediction model is greater than or equal to a preset threshold value.
Step 309(1), if the difference is greater than or equal to the predetermined threshold (the accuracy is not converged), the second prediction model is used as the first prediction model, and step 307(1) and step 308(1) are repeatedly executed.
Step 310(1), if the difference is smaller than the preset threshold (accuracy rate convergence), the first prediction model is used as the prediction model of the electricity consumption type 1, the external factor set used for training the first prediction model is determined to be the feature set of the electricity consumption type 1, and step 311(1) is executed.
And 311(1) predicting the electricity load of the electricity type 1 according to the feature set of the electricity type 1, the electricity load data of the electricity type 1 and the prediction model of the electricity type 1.
After steps 303 to 311 are executed for each power consumption type, step 312 is executed.
And step 312, predicting the total electric load according to the electric load prediction results of the electric types.
Based on the same technical concept, the embodiment of the application further provides an electrical load prediction device, and the electrical load prediction device is used for realizing the method embodiment. As shown in fig. 4, the apparatus may include: an input module 401, a prediction module 402 and an output module 403.
The input module 401 is configured to obtain power load data of a target power consumption type in a preset time period and data of multiple external factors in the preset time period.
The prediction module 402 is configured to perform causal verification on the power load data of the target power consumption type and the data of the multiple external factors, and determine that N external factors of the multiple external factors have a causal relationship with the power load of the target power consumption type; selecting M external factors from N external factors on the market in a wrapping type feature selection mode, and training a prediction model according to data of the M external factors and the power load data of the target power utilization type; and predicting the power load of the target power utilization type according to the prediction model.
And an output module 403, configured to output the prediction result.
In one possible implementation, the prediction module 402 is specifically configured to:
sorting the N external factors according to the checking result, wherein the higher the sorting is, the higher the correlation degree between the external factors and the power load of the target power consumption type is;
selecting the previous X external factors, training a first prediction model according to the X external factor data and the power load data of the target power utilization type, and estimating the accuracy of the first prediction model, wherein X is a preset constant;
training a second prediction model according to the previous M external factor data and the power load data of the target power utilization type, and estimating the accuracy of the second prediction model, wherein M is X + 1; if the value obtained by subtracting the accuracy of the first prediction model from the accuracy of the second prediction model meets the preset condition, taking the second prediction model as the first prediction model, making M equal to M +1, and repeatedly executing the step of training the second model until the value obtained by subtracting the accuracy of the first prediction model from the accuracy of the second prediction model does not meet the preset condition, or until M +1 is greater than N;
the first prediction model is used for predicting the target electricity utilization type.
In one possible implementation, the prediction module 402 is specifically configured to:
performing coordinated calibration according to the power load data of the target power utilization type and the data of the multiple external factors;
and carrying out causal verification on the data of the external factors meeting the synergistic relationship and the power load data of the target power utilization type.
In one possible implementation, the parity check is a johnson parity check; and/or, the causal check is a granger causal check.
In one possible implementation, the prediction module 402 is specifically configured to:
for each external factor, the following steps are performed:
based on the electrical load data and the external factor data, a correction model is constructed according to the following formula:
wherein, Delta YtIs a difference term of the target sequence, Δ Yt-iIs the i-order lag difference term, Γ, of the target sequenceiCoefficient matrix of difference values of i-order lag terms, ΠpIs a matrix of lag term coefficients of order p, UtIs a residual error;
according to IIpDetermines whether the external factor and the power load of the target power usage type satisfy a synergistic relationship.
In one possible implementation, the prediction module 402 is specifically configured to:
for each external factor satisfying the co-integration relationship, the following steps are performed:
constructing a time series vector autoregressive model based on the target type of the power load data and the external factor data;
time series Y to be evaluated1tAnd the external factor time series Y2tComputing residual sum of squares by least squares estimation
Time series Y to be evaluated1tComputing the sum of squares of the residuals by least squares estimation
Statistic F is calculated according to the following formula:
wherein r is zero constraint number, T is sample capacity, and p is lag period number of the vector autoregressive model;
based on x2(r) calculating rF and χ by distribution model2(r) critical value if rF.gtoreq.X2(r), the external factor has a causal relationship with the target electricity usage type.
Based on the same technical concept, the embodiment of the application further provides an electrical load prediction device, and the electrical load prediction device is used for realizing the method embodiment. As illustrated in fig. 5, the apparatus may include: an input module 501, a prediction module 502, and an output module 503.
The input module 501 is configured to obtain power consumption load data of each power consumption type in a preset time period and data of multiple external factors in the preset time period.
A prediction module 502 for, for each power usage type: determining external factors having a causal relationship with the power utilization type in the multiple external factors, selecting M external factors from the external factors having the causal relationship by adopting a wrapping type characteristic selection mode, training a prediction model according to data of the M external factors and power utilization load data of the power utilization type, and predicting the power utilization load of the power utilization type according to the prediction model; and then predicting the total power load according to the power load prediction result of each power consumption type.
And an output module 503, configured to output the total power load prediction result.
Specifically, when the prediction module 502 performs prediction for each power consumption type, the prediction manner of the prediction module 502 in the foregoing embodiment is the same as that of the prediction module.
Based on the same technical concept, embodiments of the present application also provide a computer-readable storage medium storing computer instructions, which, when executed on a computer, cause the computer to execute the above electrical load prediction method.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.
Claims (10)
1. An electrical load prediction method, comprising:
acquiring power load data of a target power utilization type in a preset time period and data of various external factors in the preset time period;
carrying out causal verification on the power load data of the target power utilization type and the multiple external factor data, and determining that N external factors in the multiple external factors have causal relation with the power load of the target power utilization type;
selecting M external factors from the N external factors in a wrapping type feature selection mode, and training a prediction model according to the data of the M external factors and the power load data of the target power utilization type;
and predicting the power load of the target power utilization type according to the prediction model, and outputting a prediction result.
2. The method of claim 1, wherein the step of selecting the M external factors from the N external factors in a wrapping-type feature selection manner, and the step of training a prediction model according to data of the M external factors and power load data of the target power consumption type comprises:
sorting the N external factors according to the checking result, wherein the higher the sorting is, the higher the correlation degree between the external factors and the power load of the target power consumption type is;
selecting the previous X external factors, training a first prediction model according to the X external factor data and the power load data of the target power utilization type, and estimating the accuracy of the first prediction model, wherein X is a preset constant;
training a second prediction model according to the previous M external factor data and the power load data of the target power utilization type, and estimating the accuracy of the second prediction model, wherein M is X + 1; if the value obtained by subtracting the accuracy of the first prediction model from the accuracy of the second prediction model meets the preset condition, taking the second prediction model as the first prediction model, making M equal to M +1, and repeatedly executing the step of training the second model until the value obtained by subtracting the accuracy of the first prediction model from the accuracy of the second prediction model does not meet the preset condition, or until M +1 is greater than N;
the first prediction model is used for predicting the target electricity utilization type.
3. The method of claim 1, wherein said causally verifying the power load data of the target power usage type and the plurality of external factor data comprises:
performing coordinated calibration according to the power load data of the target power utilization type and the data of the multiple external factors;
and carrying out causal verification on the data of the external factors meeting the synergistic relationship and the power load data of the target power utilization type.
4. The method of claim 3, wherein the integrity check is a Johnson integrity check; and/or
The causal test is a granger causal test.
5. The method according to claim 4, wherein said conducting a Johnson coordination check according to the power load data of the target power usage type and each of the external factor data, comprises:
for each external factor, the following steps are performed:
based on the electrical load data and the external factor data, a correction model is constructed according to the following formula:
wherein, Delta YtIs a difference term of the target sequence, Δ Yt-iIs the i-order lag difference term, Γ, of the target sequenceiIs a coefficient matrix of i-order lag term difference values, < pi >pIs a matrix of lag term coefficients of order p, UtIs a residual error;
according to the IIpDetermines whether the external factor and the power load of the target power usage type satisfy a synergistic relationship.
6. The method of claim 4, wherein performing a granger causal check on the data of the external factors satisfying the synergistic relationship and the power load data of the target power usage type comprises:
for each external factor satisfying the co-integration relationship, the following steps are performed:
constructing a time series vector autoregressive model based on the target type of the power load data and the external factor data;
time series Y to be evaluated1tAnd the external factor time series Y2tComputing residual sum of squares by least squares estimation
Time series Y to be evaluated1tCalculating the residual squared by least squares estimationAnd
statistic F is calculated according to the following formula:
wherein r is zero constraint number, T is sample capacity, and p is lag period number of the vector autoregressive model;
based on x2(r) calculating rF and χ by distribution model2(r) critical value if rF.gtoreq.X2(r), the external factor has a causal relationship with the target electricity usage type.
7. An electrical load prediction method, comprising:
performing power consumption load prediction according to the power consumption load prediction method of any one of claims 1 to 6 for each power consumption type;
and predicting the total power load according to the power load prediction result of each power consumption type, and outputting the prediction result.
8. An electrical load prediction apparatus, comprising:
the input module is used for acquiring power load data of a target power utilization type in a preset time period and data of various external factors in the preset time period;
the prediction module is used for carrying out causal verification on the power load data of the target power utilization type and the multiple external factor data and determining that N external factors in the multiple external factors have causal relationship with the power load of the target power utilization type; selecting M external factors from the N external factors in a wrapping type feature selection mode, and training a prediction model according to the data of the M external factors and the power load data of the target power utilization type; predicting the power load of the target power utilization type according to the prediction model;
and the output module is used for outputting the prediction result.
9. An electrical load prediction apparatus, comprising:
the input module is used for acquiring power load data of each power utilization type in a preset time period and various external factor data in the preset time period;
a prediction module for performing a power consumption load prediction according to the power consumption load prediction method of any one of claims 1 to 6 for each power consumption type; predicting the total power load according to the power load prediction result of each power consumption type;
and the output module is used for outputting the prediction result.
10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911178565.8A CN110942195A (en) | 2019-11-27 | 2019-11-27 | Power load prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911178565.8A CN110942195A (en) | 2019-11-27 | 2019-11-27 | Power load prediction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110942195A true CN110942195A (en) | 2020-03-31 |
Family
ID=69908180
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911178565.8A Pending CN110942195A (en) | 2019-11-27 | 2019-11-27 | Power load prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110942195A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113435923A (en) * | 2021-06-15 | 2021-09-24 | 北京百度网讯科技有限公司 | Power consumption prediction method and device and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104834975A (en) * | 2015-05-13 | 2015-08-12 | 国家电网公司 | Power network load factor prediction method based on intelligent algorithm optimization combination |
| US20190081476A1 (en) * | 2017-09-12 | 2019-03-14 | Sas Institute Inc. | Electric power grid supply and load prediction |
| CN110222877A (en) * | 2019-05-17 | 2019-09-10 | 华中科技大学 | A kind of load prediction system and load forecasting method based on customized neural network |
| CN110503256A (en) * | 2019-08-14 | 2019-11-26 | 北京国网信通埃森哲信息技术有限公司 | Short-term load forecasting method and system based on big data technology |
-
2019
- 2019-11-27 CN CN201911178565.8A patent/CN110942195A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104834975A (en) * | 2015-05-13 | 2015-08-12 | 国家电网公司 | Power network load factor prediction method based on intelligent algorithm optimization combination |
| US20190081476A1 (en) * | 2017-09-12 | 2019-03-14 | Sas Institute Inc. | Electric power grid supply and load prediction |
| CN110222877A (en) * | 2019-05-17 | 2019-09-10 | 华中科技大学 | A kind of load prediction system and load forecasting method based on customized neural network |
| CN110503256A (en) * | 2019-08-14 | 2019-11-26 | 北京国网信通埃森哲信息技术有限公司 | Short-term load forecasting method and system based on big data technology |
Non-Patent Citations (2)
| Title |
|---|
| 刘俊 等: "基于协整—格兰杰因果检验和季节分解的中期负荷预测", 《电力系统自动化》 * |
| 胡忠义: "《基于计算智能技术的电力负荷预测理论及应用》", 31 January 2019, 武汉大学出版社 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113435923A (en) * | 2021-06-15 | 2021-09-24 | 北京百度网讯科技有限公司 | Power consumption prediction method and device and electronic equipment |
| CN113435923B (en) * | 2021-06-15 | 2022-09-27 | 北京百度网讯科技有限公司 | Power consumption prediction method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112734128B (en) | 7-day power load peak prediction method based on optimized RBF | |
| CN112215442B (en) | Method, system, device and medium for predicting short-term load of power system | |
| CN111680841B (en) | Short-term load prediction method, system and terminal equipment based on principal component analysis | |
| CN110837915B (en) | Low-voltage load point prediction and probability prediction method for power system based on hybrid integrated deep learning | |
| CN112365056A (en) | Electrical load joint prediction method and device, terminal and storage medium | |
| CN113822481A (en) | Comprehensive energy load prediction method based on multi-task learning strategy and deep learning | |
| CN115470862A (en) | Dynamic self-adaptive load prediction model combination method | |
| CN112381315A (en) | LS-SVM intelligent platform area load prediction method and system based on PSO optimization | |
| CN117674119A (en) | Power grid operation risk assessment method, device, computer equipment and storage medium | |
| Yan et al. | Short-term load forecasting of smart grid based on load spatial-temporal distribution | |
| CN115619028A (en) | Clustering algorithm fusion-based power load accurate prediction method | |
| CN111680712A (en) | Transformer oil temperature prediction method, device and system based on similar moments in the day | |
| CN116823008A (en) | Park energy utilization efficiency evaluation method, system, equipment and storage medium | |
| CN118536646A (en) | Prediction method, system and equipment for long-term dynamic carbon emission factor in region | |
| Zuo | Integrated forecasting models based on LSTM and TCN for short-term electricity load forecasting | |
| CN110942195A (en) | Power load prediction method and device | |
| CN117742160A (en) | Artificial intelligence-based carbon emission optimization control method and system | |
| CN113887809A (en) | Power distribution network supply and demand balance method, system, medium and computing equipment under double-carbon target | |
| CN115528750B (en) | Power grid safety and stability oriented data model hybrid drive unit combination method | |
| Daraghmi et al. | Accurate and time‐efficient negative binomial linear model for electric load forecasting in IoE | |
| CN114759579A (en) | Power grid active power optimization control system, method and medium based on data driving | |
| CN110222879B (en) | System-bus load prediction coordination method and device considering confidence interval | |
| CN113962751A (en) | Retail price prediction method and device for electric power package and storage medium | |
| CN114330865A (en) | Power grid reserve capacity prediction method and system, computer equipment and storage medium | |
| CN113887816A (en) | Intelligent prediction method, device and equipment for social reserve of household appliances |
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: 20200331 |