CN117543544A - Load prediction method, device, equipment and storage medium - Google Patents
Load prediction method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN117543544A CN117543544A CN202311443662.1A CN202311443662A CN117543544A CN 117543544 A CN117543544 A CN 117543544A CN 202311443662 A CN202311443662 A CN 202311443662A CN 117543544 A CN117543544 A CN 117543544A
- Authority
- CN
- China
- Prior art keywords
- data
- component
- influence factor
- sequence
- power load
- 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 53
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 41
- 238000012545 processing Methods 0.000 claims abstract description 19
- 238000005516 engineering process Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 31
- 238000004590 computer program Methods 0.000 claims description 27
- 230000015654 memory Effects 0.000 claims description 18
- 238000012360 testing method Methods 0.000 claims description 12
- 230000005611 electricity Effects 0.000 claims description 11
- 230000001186 cumulative effect Effects 0.000 claims description 10
- 230000001133 acceleration Effects 0.000 claims description 9
- 238000003062 neural network model Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 4
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims description 2
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims description 2
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 15
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000009825 accumulation Methods 0.000 description 4
- 230000009467 reduction Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000004069 differentiation Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000001556 precipitation Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Water Supply & Treatment (AREA)
- Operations Research (AREA)
- Power Engineering (AREA)
- General Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a load prediction method, a load prediction device, load prediction equipment and a storage medium, wherein the load prediction method comprises the steps of obtaining historical power load data and influence factor data of each subarea; respectively processing the historical power load data according to an empirical mode decomposition technology to obtain corresponding component sequence data; determining main influence factor data of the component sequence data based on the similarity between the influence factor data and the component sequence data; constructing a corresponding component prediction model by using the main influence factor data and the component sequence data; and determining the power load prediction result of the target area according to the component prediction results of all the component prediction models. According to the load prediction method provided by the invention, the historical power load data is subjected to empirical mode decomposition to obtain a plurality of component sequences, the main influence factors corresponding to each component sequence are found, the component prediction model corresponding to each component sequence is obtained, and the power load prediction models are obtained by fusing all the component prediction models, so that the power load can be predicted more accurately.
Description
Technical Field
The present invention relates to the field of power load technologies, and in particular, to a load prediction method, apparatus, device, and storage medium.
Background
The power load prediction is important to the power company to effectively manage demand response and is used for guiding the power company to schedule the generator set and manage energy distribution. The power load variation is affected by multiple factors, with the climate being most pronounced. The power load has the characteristics of strong non-stationarity and nonlinearity, but has the characteristics of periodicity locally, so that the load prediction is challenged. However, for social environmental impact, the population number in different areas is different, and the power load is also changed due to the difference of the economic development degree and the power demand. Accurate load prediction is of great significance in improving power planning and grid stability. However, because of numerous factors affecting the load and complex interactions between the influencing factors, it is particularly difficult to reasonably construct a load prediction model. The traditional load prediction method has the defect that complex relations between factors such as weather, social environment and the like and loads are difficult to capture, so that influences of meteorological and social environment factors on power loads are not considered, and prediction accuracy is low.
In the prior art, for example, a "training method and apparatus for a power load prediction model and a power load prediction method" disclosed in chinese patent CN115293326a, the power load prediction model includes a full convolutional neural network FCN and a long short term memory network LSTM, and the training method includes: acquiring a training data set of a current training period, wherein the training data set comprises time sequence data of historical load; inputting the time sequence data in the training data set into a power load prediction model of the current training period to obtain a power load prediction value of the current training period; determining a loss function of the current training period based on the predicted value of the power load of the current training period and the true value of the power load of the current training period; and adjusting parameters of the power load prediction model of the current training period according to the loss function of the current training period. According to the technology, after the power load prediction model is obtained through the training of the historical power load data, the load is predicted, and the influence relationship between the power load and weather and social environment factors is not considered, so that the prediction accuracy is not high.
Disclosure of Invention
The invention provides a load prediction method, a device, equipment and a storage medium, which are used for solving the technical problem that the prediction accuracy is low because the influence of meteorological factors and social environment factors on the power load is not considered in the existing power load prediction, and the power load can be predicted more accurately by carrying out empirical mode decomposition on historical power load data to obtain a plurality of component sequences, finding out main influence factors corresponding to each component sequence, obtaining a component prediction model corresponding to each component sequence and fusing all component prediction models to obtain the power load prediction model.
In order to solve the above technical problems, an embodiment of the present invention provides a load prediction method, including:
respectively acquiring historical power load data and influence factor data corresponding to each subarea in a target area, wherein the types of the influence factor data at least comprise daily maximum temperature, daily minimum temperature, single month accumulated high Wen Tianshu, single month accumulated low temperature days, town rate, resident population, economic acceleration and industrial electricity utilization structure;
according to an empirical mode decomposition technology, processing each historical power load data to obtain corresponding component sequence data;
determining main influence factor data corresponding to each component sequence data based on the similarity between each influence factor data and each component sequence data;
constructing a corresponding component prediction model by using the main influence factor data and the component sequence data;
and determining the power load prediction result of the target area according to the component prediction results of all the component prediction models.
As one preferable solution, the processing, according to an empirical mode decomposition technique, each historical power load data to obtain corresponding component sequence data specifically includes:
adding Gaussian white noise for preset times to the power load data to be decomposed to form a plurality of subsequences to be decomposed;
when empirical mode decomposition is performed, performing empirical mode decomposition on the subsequence to be decomposed, taking the average value of the obtained first modal components as a first modal component sequence of power load data to be decomposed, and calculating to obtain a corresponding first residual signal;
after adding a first modal component of Gaussian white noise into the first residual signal, continuing empirical mode decomposition, taking the average value of the obtained second modal component as a second modal component sequence of power load data to be decomposed, and calculating to obtain a corresponding second residual signal;
repeating the empirical mode decomposition until the empirical mode decomposition stopping condition is satisfied.
As one preferable solution, the determining, based on the similarity between each of the influence factor data and each of the component sequence data, main influence factor data corresponding to each of the component sequence data specifically includes:
sequencing the influence factor data based on the time axis sequence to obtain a corresponding influence factor sequence;
carrying out standardization treatment on the influence factor sequence;
respectively calculating the similarity between the component sequence data and the influence factor sequences after each normalization processing;
and determining the influence factor sequence corresponding to the similarity result being larger than a preset similarity threshold value as main influence factor data.
As one preferable solution, the calculating the similarity between the component sequence data and the influence factor sequence after each normalization processing specifically includes:
the similarity between the component sequence and the influencing factor sequence is calculated according to the following cumulative distance formula:
γ(i,j)＝d(f i ,p j )+min{γ(i-1,j-1),γ(i-1,j),γ(i,j+1)}
wherein, gamma is the accumulated distance, f i For the ith normalized component value in the component sequence F, the component sequence f= { F 1 ,f 2 ,…,f i ,…,f n }，p j To influence the j-th normalized component value in the factor sequence P, the factor sequence p= { P is influenced 1 ,p 2 ,…,p j ,…,p m }，d(f i ,p j ) For the distance of each component in the sequence of components F and the sequence of influencing factors P from each other.
As one preferable scheme, the constructing a corresponding component prediction model according to the main influence factor data and the component sequence data specifically includes:
dividing a set formed by the component sequence data and the main influence factor data into a training set and a testing set;
and inputting the samples of the training set into a neural network model for training, taking the mean square error as a training index, selecting the parameters corresponding to the minimum mean square error to generate a component prediction model, and evaluating through the test set.
As one preferable scheme, the evaluating by the test set specifically includes:
the mean absolute error percentage and the mean square error calculated according to the following formula are evaluated:
wherein MAPE is the average absolute error percentage, MSE is the mean square error, n is the number of data, y i As the real data of the data to be processed,is predictive data.
As one preferable mode, the determining the power load prediction result of the target area according to the component prediction results of all the component prediction models specifically includes:
obtaining a component prediction result corresponding to each component prediction model;
and carrying out summation processing on all the component prediction results to obtain the power load prediction result of the target area.
Another embodiment of the present invention provides a load predicting apparatus including:
the data acquisition module is used for respectively acquiring historical power load data and influence factor data corresponding to each subarea in the target area, wherein the types of the influence factor data at least comprise daily maximum temperature, daily minimum temperature, single month accumulated high Wen Tianshu, single month accumulated low temperature days, town ratio, resident population, economic acceleration and industrial electricity utilization structure;
the decomposition module is used for respectively processing each historical power load data according to an empirical mode decomposition technology to obtain corresponding component sequence data;
the main influence factor module is used for determining main influence factor data corresponding to each component sequence data based on the similarity between each influence factor data and each component sequence data;
the model construction module is used for constructing a corresponding component prediction model by the main influence factor data and the component sequence data;
and the load prediction module is used for determining the power load prediction result of the target area according to the component prediction results of all the component prediction models.
Yet another embodiment of the present invention provides a load prediction device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the load prediction method as described above when executing the computer program.
Still another embodiment of the present invention provides a computer readable storage medium storing a computer program, where the load prediction method described above is implemented when the computer program is executed by a device in which the computer readable storage medium is located.
Compared with the prior art, the embodiment of the invention has the beneficial effects that at least one of the following points is adopted:
(1) In the invention, the fact that the power load data is influenced by meteorological data and social environment factors is considered to generate differences due to different geographic positions, so that each subarea is obtained according to geographic position classification, and a corresponding power load prediction model of each subarea is obtained according to the geographic position of each subarea for load prediction, and compared with the large-scale integral power load prediction, the prediction result according to the geographic position of the geographic city differentiation is more accurate;
(2) In the invention, considering that the influence factors on the power load data are various and have different relevance, only one model is used for predicting the power load data by considering the influence factors, the relation is complex and difficult to clear, so that the power load data are decomposed into different component sequences, each component sequence also has the influence factors with corresponding larger influence degree, the similarity of the component sequences and the influence factor sequences is calculated, and the influence factors meeting the conditions are selected as main influence factors capable of influencing the component sequences, thereby improving the accuracy of subsequent prediction;
(3) The method comprises the steps of carrying out empirical mode decomposition on historical power load data to obtain a plurality of component sequences, finding out main influence factors matched with the component sequences for each component sequence, so as to obtain component prediction models corresponding to each component sequence, wherein each component prediction model comprises different influence factors on the power load, and the power load prediction models obtained by fusing all component prediction models can be used for accurately predicting the power load under the condition of considering weather and social environment factors.
Drawings
FIG. 1 is a flow chart of a load prediction method in one embodiment of the invention;
FIG. 2 is a block diagram of a load predicting device in one embodiment of the present invention;
FIG. 3 is a block diagram of a load predicting device in one embodiment of the invention;
reference numerals:
11, a data acquisition module; 12. a decomposition module; 13. a main influencing factor module; 14. a model building module; 15. a load prediction module; 21. a processor; 22. a memory; .
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention, and the purpose of these embodiments is to provide a more thorough and complete disclosure of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of this application, the terms "first," "second," "third," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first", "a second", "a third", etc. may explicitly or implicitly include one or more such feature. In the description of the present application, unless otherwise indicated, the meaning of "a plurality" is two or more.
In the description of the present application, it should be noted that, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The terms "vertical," "horizontal," "left," "right," "upper," "lower," and the like are used herein for descriptive purposes only and not to indicate or imply that the apparatus or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore should not be construed as limiting the invention. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items. The specific meaning of the terms in this application will be understood by those of ordinary skill in the art in a specific context.
In the description of the present application, it should be noted that all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs unless defined otherwise. The terminology used in the description of the present invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention, as the particular meaning of the terms described above in this application will be understood to those of ordinary skill in the art in the specific context.
An embodiment of the present invention provides a load prediction method, specifically, referring to fig. 1, fig. 1 is a schematic flow chart of a load prediction method in one embodiment of the present invention, which specifically includes steps S1 to S5:
s1, respectively acquiring historical power load data and influence factor data corresponding to each subarea in a target area, wherein the types of the influence factor data at least comprise daily maximum temperature, daily minimum temperature, single month accumulated high Wen Tianshu, single month accumulated low temperature days, town rate, resident population, economic acceleration and industrial electricity utilization structure;
s2, respectively processing each historical power load data according to an empirical mode decomposition technology to obtain corresponding component sequence data;
s3, determining main influence factor data corresponding to each component sequence data based on the similarity between each influence factor data and each component sequence data;
s4, constructing a corresponding component prediction model by using the main influence factor data and the component sequence data;
s5, determining a power load prediction result of the target area according to the component prediction results of all the component prediction models.
It should be noted that, in the embodiment of the present invention, different sub-areas are obtained by dividing according to geographic locations, so as to obtain respective historical power load data and influence factor data of different geographic locations. In addition, regarding the influence factor data, the embodiment of the invention takes the accumulation of high Wen Tianshu and low temperature days in a month as examples, and specifically comprises the following steps:
the setting process of the high temperature lower limit value in the single month integrated high Wen Tianshu includes: for the highest temperature sequence samples per day { t } 1 ,t 2 ,…,t n Sequence of constructs:when t i >t j J=1, 2, …, r at i i Accumulate 1, otherwise 0; and defining statistical variables:
wherein the method comprises the steps ofSum sigma(s) l ) S are respectively l Mean and variance of (a); repeating the above steps in reverse order to obtain UB l ＝-UF l L=n, n-1, …,1; with UF l And UB l The temperature corresponding to the intersection of the two curves is taken as the high temperature lower limit value.
The setting process of the low temperature upper limit value in the single month accumulated low temperature days is similar to the setting process of the high temperature lower limit value, and the low temperature upper limit threshold value is obtained through analysis of the daily lowest temperature sequence sample.
The specific types of influencing factors in the embodiment of the invention can include: day maximum temperature, month maximum temperature, day minimum temperature, month minimum temperature, and month cumulative height Wen Tianshu (month is higher than high temperature lower limit value t) High temperature Days of (2) above t High temperature Days +1 up to above t High temperature Days of +10, i.e. 1 degree celsiusStep size statistics is performed on the accumulated days of the temperature above the high temperature lower limit value), and the accumulated low temperature days of a month (the current month is lower than the low temperature upper limit value t) Low temperature Days below t Low temperature Days-1 up to below t Low temperature -10 days, i.e. counting cumulative days at a temperature below the upper limit of low temperature with a step size of 1 degree celsius), humidity, wind power, precipitation, township rate, month GDP acceleration, quarter GDP acceleration, resident population, first industry electricity utilization ratio, second industry electricity utilization ratio, third industry electricity utilization ratio, resident electricity utilization ratio, etc.
Further, in the above embodiment, empirical mode decomposition is performed on the historical power load data to obtain corresponding component sequence data, specifically, K modal component sequences and one residual component sequence.
The process of empirical mode decomposition of historical electrical load data includes:
adding Gaussian white noise with the mean value of 0K times to power load data x (t) to be decomposed to form K subsequences to be decomposed;
x i (t)＝x(t)+εδ i (t)
wherein epsilon is the weight coefficient of Gaussian white noise, delta i (t) Gaussian white noise added for the ith sequence to be decomposed, i.e. [1, K]。
The first modal component obtained by EMD decomposition of K subsequences to be decomposed is averaged to be used as the 1 st modal component sequence IMF of x (t) 1 (t) and calculating the 1 st residual signal r 1 (t)；
r 1 (t)＝x(t)-IMF 1 (t)
Wherein the method comprises the steps ofRepresenting a first modal component obtained by EMD decomposition of the ith sub-sequence to be decomposed.
At the j-1 th residual signal r j-1 Adding Gaussian white to (t)The j-1 mode component of the noise is continuously subjected to EMD decomposition to obtain the j mode component sequence IMF of x (t) j (t) and calculating the j-th residual signal r j (t)；
r j (t)＝r j-1 (t)-IMF j (t)
Wherein E is j-1 (. Cndot.) represents the j-1 th modal component, ε, of the sequence after EMD decomposition j-1 The weight coefficient indicating the addition of noise to the j-1 th residual signal.
Repeating the steps until the EMD decomposition stopping condition is met, and completing the decomposition to obtain K modal component sequences and a residual component sequence if the K-th decomposed residual signal is a monotonic signal.
Further, in the above embodiment, the corresponding main influence factor data (i.e., influence factor principal component) is matched for each component sequence, specifically:
the influence factor data is sequenced according to a time axis to generate an influence factor sequence, and d influence factor sequences can be obtained when d influence factors are used in total; carrying out standardization processing on the influence factor sequence and the component sequence; and calculating the similarity of the component sequence and the influence factor sequence, and matching the influence factor sequence with the similarity larger than a preset similarity threshold value into main influence factor data (influence factor main components) corresponding to the component sequence.
In the above embodiment, the process of calculating the similarity of the component sequence and the influence factor sequence includes:
for the component sequence f= { F 1 ,f 2 ,…,f i ,…,f n The sequence of } and influencing factors p= { P 1 ,p 2 ,…,p j ,…,p m }，f i And p j Representing normalized values corresponding to the respective sequences; calculating the distance d (F) between each of the components F and P i ,p j ) The method comprises the steps of carrying out a first treatment on the surface of the According to the cumulative distance formula γ (i, j) =d (f) i ,p j ) +min { γ (i-1, j-1), γ (i-1, j), γ (i, j+1) } calculationThe minimum cumulative distance between F and P is taken as the similarity of the component sequence and the influence factor sequence.
Calculating to obtain similarity results of the kth component sequence and all influence factor sequencesAnd selecting all similarity results larger than a similarity threshold value from the similarity results, wherein the corresponding influence factors are the influence factor main components of the kth component sequence, so that the influence factor main components corresponding to all the component sequences are obtained by matching.
Further, in the above embodiment, constructing the corresponding component prediction model with the main influence factor data and the component sequence data includes:
the method comprises the steps of dividing a set consisting of a component sequence and a main component of an influence factor into a training set and a testing set, inputting samples of the training set into a neural network model for training, taking mean square error as a training index, selecting parameters corresponding to the minimum mean square error to generate a component prediction model, and evaluating through the testing set.
Evaluation by the test set includes at least calculating an average absolute percentage and calculating a mean square error:
where n is a number of data, y i As the real data of the data to be processed,is predictive data.
And then, fusing all the component prediction models to obtain a power load prediction structure, and realizing load prediction.
Specifically, the load prediction process includes: and respectively predicting the K+1 component prediction models to obtain K+1 component prediction results, and summing all the component prediction results to obtain the power load prediction result of the target area.
In the invention, the power load data is influenced by meteorological data and social environment factors and also can be different due to different geographic positions; firstly classifying according to geographic positions, wherein each position area corresponds to historical power load and influence factor data of the position area, and obtaining a corresponding power load prediction model for each geographic position to perform load prediction; compared with the prediction of a large-scale overall power load, the prediction result of the city differentiation according to the geographic position is more accurate; therefore, firstly, the historical power load data and influence factors are classified according to geographic positions (such as different cities or regions), the historical power load data are decomposed to obtain different component sequences, the data of each component sequence are actually related to a plurality of different influence factors, therefore, a plurality of components with the largest influence degree are matched to the corresponding component sequences as main components of the influence factors, so that a component prediction model is trained and obtained, the influence factors of the different component prediction models with different weights are different, and therefore, the power load prediction model obtained after fusion can also be used for carrying out power load prediction of the corresponding region based on a plurality of influence factors, and the result is more accurate.
The weather data comprises the highest daily temperature, the lowest daily temperature, the accumulated high per month Wen Tianshu and the accumulated low temperature days per month, the highest daily temperature judgment standard of a certain region is the median of the highest temperatures detected by all weather stations of the region, and the lowest temperature is the median of the lowest temperatures detected by all weather stations of the region; meteorological factors may also include humidity, wind, precipitation, etc.; in addition to the influence of meteorological factors, the power load is greatly affected by different economic development levels, different town levels, population differences and the like caused by regional differences.
The invention adopts the empirical mode decomposition of a fully self-adaptive noise set, adds the mode component containing auxiliary noise after EMD decomposition, carries out ensemble average calculation in the first stage of decomposition to obtain the mode component IMF of the first order, and then repeatedly carries out the operation of adding auxiliary noise and decomposing on the residual part to finally obtain K mode component sequences and a residual component sequence; thereby decomposing the electrical load data into k+1 different portions of load data.
In the invention, as the influence factors on the power load data are various and have different correlations, only one model is used for considering the influence factors to predict the power load data, and the relation is complex and difficult to clear; the method comprises the steps of firstly decomposing power load data into different component sequences, wherein each component sequence also has corresponding influence factors with larger influence degree, calculating the similarity of the component sequences and the influence factor sequences, and selecting the influence factors meeting the conditions as the main components of the influence factors capable of influencing the component sequences; each component sequence has a corresponding influence factor principal component that is not identical.
The acquisition time intervals of the component sequences and the influence factor sequences may be different, and the time lengths may also be different, so that the input sequences are required to be distorted by extending and shortening the sequences, so that the two sequences are aligned in time, the corresponding relation between the test template and the reference template is described by using a regular function meeting a certain condition, and an optimal path is selected to solve the minimum accumulated distance between the two sequences to be used as a similarity value; the distance between the two components may be euclidean distance or manhattan distance, etc.
The purpose of the present invention is to reflect the influence of the high temperature accumulation effect on the power load data, which is a factor of the accumulation of the high temperature Wen Tianshu per month, so that the high temperature lower limit t High temperature As the judgment basis for the high temperature accumulation effect, the corresponding index in the influence factors of the invention can have a temperature of one month greater than t High temperature The number of days of (a), the temperature of a month is greater than t High temperature Days of +1, etc.; upper limit t of low temperature in the number of days of accumulated low temperature for one month Low temperature The calculation mode is similar to the high temperature lower limit value, and the temperature for one month can also be set to be lower than t Low temperature The number of days of (a), the temperature of a month is lower than t Low temperature Days of-1.
The neural network model selected by the component prediction model can be an LSTM long-term memory model, or a GRU model and the like, which is trained through a training set and verified through a testing set, and the neural network model is retrained if the verification is not passed, and the neural network model is used as a final component prediction model if the verification is passed; in the training and testing process of the component prediction model, the data of the component sequence and the main components of the influencing factors are all the data before the standardization processing, and the purpose of the standardization of the data is just to match and facilitate the calculation of the similarity.
When the adopted neural network model is an LSTM neural network model, the system comprises a plurality of LSTM units, and each LSTM unit is provided with an input gate, a forget gate and an output gate and is used for protecting and controlling the states of the units:
forgetting door f t Determining that certain information needs to be forgotten from the current unit state, and outputting information h of the previous unit t-1 With newly entered information x t Outputting a number between 0 and 1 through a sigmoid activation function, wherein 1 represents that the information is completely reserved, and 0 represents that the information is completely forgotten:
f t ＝σ(W f ·[h t-1 ,x t ]+b f )
input gate i t Determining which information needs to be stored in the current unit, and outputting information h of the previous unit t-1 With newly entered information x t Creating a new cell state value vector through a function tanhWith sigmoid activation function to update cell state C t ：
i t ＝σ(W i ·[h t-1 ,x t ]+b i )
Output door o t Determining the output information of the current cell state, processing the cell state by a tanh function, and then combining withMultiplying the output values of the sigmoid function to finally determine output information h t ：
o t ＝σ(W o ·[h t-1 ,x t ]+b o )
h t ＝o t ·tanh(C t )
Wherein W is f 、W i 、W C And W is o Weight matrix of forgetting gate, input gate, cell state and output gate, b f 、b i 、b C And b o Respectively corresponding bias matrices.
In the invention, when model training is performed, the condition that the value of the mean square error gradually decreases but does not reach the minimum value all the time can exist, a reduction threshold value can be set at the moment, and when the value of the mean square error reduction is smaller than the reduction threshold value, training can be considered to be completed; training with other indicators may also be determined by setting a reduction threshold.
The component prediction model comprises component prediction models of K modal component sequences and component prediction models of one residual component sequence, corresponding load predictions are respectively carried out on different component prediction models according to the main components of the corresponding influence factors of each component prediction model, and the prediction results are summed to obtain a final power load prediction result, so that the prediction result is more accurate.
Referring to fig. 2, fig. 2 shows a block diagram of a load predicting device according to one embodiment of the present invention, which includes:
the data acquisition module 11 is configured to acquire historical power load data and influence factor data corresponding to each sub-region in the target region, where the types of the influence factor data at least include a highest daily temperature, a lowest daily temperature, a cumulative high per month Wen Tianshu, a cumulative low-temperature days per month, a town ratio, a resident population, an economic acceleration rate, and an industrial electricity utilization structure;
the decomposition module 12 is configured to process each of the historical power load data according to an empirical mode decomposition technique, so as to obtain corresponding component sequence data;
a main influencing factor module 13, configured to determine main influencing factor data corresponding to each component sequence data based on a similarity between each influencing factor data and each component sequence data;
a model construction module 14, configured to construct a corresponding component prediction model according to the main influence factor data and the component sequence data;
and the load prediction module 15 is used for determining the power load prediction result of the target area according to the component prediction results of all the component prediction models.
Referring to fig. 3, which is a block diagram of a load prediction device provided by an embodiment of the present invention, the load prediction device provided by the embodiment of the present invention includes a processor 21, a memory 22, and a computer program stored in the memory 22 and configured to be executed by the processor 21, where the processor 21 implements steps in the embodiment of the load prediction method described above, such as steps S1 to S5 described in fig. 1, when executing the computer program; alternatively, the processor 21 may implement the functions of the modules in the above-described device embodiments, such as the data acquisition module 11, when executing the computer program.
Illustratively, the computer program may be split into one or more modules that are stored in the memory 22 and executed by the processor 21 to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions for describing the execution of the computer program in the load predicting device. For example, the computer program may be divided into a data acquisition module 11, a decomposition module 12, a main influencing factor module 13, a model construction module 14, and a load prediction module 15, each of which functions as follows:
the data acquisition module 11 is configured to acquire historical power load data and influence factor data corresponding to each sub-region in the target region, where the types of the influence factor data at least include a highest daily temperature, a lowest daily temperature, a cumulative high per month Wen Tianshu, a cumulative low-temperature days per month, a town ratio, a resident population, an economic acceleration rate, and an industrial electricity utilization structure;
the decomposition module 12 is configured to process each of the historical power load data according to an empirical mode decomposition technique, so as to obtain corresponding component sequence data;
a main influencing factor module 13, configured to determine main influencing factor data corresponding to each component sequence data based on a similarity between each influencing factor data and each component sequence data;
a model construction module 14, configured to construct a corresponding component prediction model according to the main influence factor data and the component sequence data;
and the load prediction module 15 is used for determining the power load prediction result of the target area according to the component prediction results of all the component prediction models.
The load predicting device may include, but is not limited to, a processor 21, a memory 22. It will be appreciated by those skilled in the art that the schematic diagram is merely an example of a load predicting device, and does not constitute a limitation of the load predicting device, and may include more or less components than illustrated, or may combine certain components, or different components, e.g., the load predicting device may further include an input-output device, a network access device, a bus, etc.
The processor 21 may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, and the processor 21 is a control center of the load predicting apparatus, and connects the respective parts of the entire load predicting apparatus using various interfaces and lines.
The memory 22 may be used to store the computer program and/or module, and the processor 21 may implement various functions of the load predicting device by executing or executing the computer program and/or module stored in the memory 22 and invoking data stored in the memory 22. The memory 22 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program (such as a sound playing function, an image playing function, etc.) required for at least one function, and the like; the storage data area may store data (such as audio data, phonebook, etc.) created according to the use of the handset, etc. In addition, the memory 22 may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart Media Card (SMC), secure Digital (SD) Card, flash Card (Flash Card), at least one disk storage device, flash memory device, or other volatile solid-state storage device.
Wherein the load predicting device integrated modules may be stored in a computer readable storage medium if implemented in the form of software functional units and sold or used as a stand alone product. Based on such understanding, the present invention may implement all or part of the flow of the method of the above embodiment, or may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, a software distribution medium, and so forth.
Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program stored on a computer readable storage medium, which when executed may comprise the steps of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), or the like.
Accordingly, an embodiment of the present invention provides a computer readable storage medium, where the computer readable storage medium includes a stored computer program, where when the computer program runs, the computer readable storage medium is controlled to execute steps in the load prediction method according to the foregoing embodiment, for example, steps S1 to S5 described in fig. 1.
The load prediction method, the load prediction device, the load prediction equipment and the storage medium provided by the embodiment of the invention have the beneficial effects that at least one point of the following is:
(1) In the invention, the fact that the power load data is influenced by meteorological data and social environment factors is considered to generate differences due to different geographic positions, so that each subarea is obtained according to geographic position classification, and a corresponding power load prediction model of each subarea is obtained according to the geographic position of each subarea for load prediction, and compared with the large-scale integral power load prediction, the prediction result according to the geographic position of the geographic city differentiation is more accurate;
(2) In the invention, considering that the influence factors on the power load data are various and have different relevance, only one model is used for predicting the power load data by considering the influence factors, the relation is complex and difficult to clear, so that the power load data are decomposed into different component sequences, each component sequence also has the influence factors with corresponding larger influence degree, the similarity of the component sequences and the influence factor sequences is calculated, and the influence factors meeting the conditions are selected as main influence factors capable of influencing the component sequences, thereby improving the accuracy of subsequent prediction;
(3) The method comprises the steps of carrying out empirical mode decomposition on historical power load data to obtain a plurality of component sequences, finding out main influence factors matched with the component sequences for each component sequence, so as to obtain component prediction models corresponding to each component sequence, wherein each component prediction model comprises different influence factors on the power load, and the power load prediction models obtained by fusing all component prediction models can be used for accurately predicting the power load under the condition of considering weather and social environment factors.
The foregoing examples illustrate only a few embodiments of the invention and are described in detail herein without thereby limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Claims (10)
1. A load prediction method, comprising:
respectively acquiring historical power load data and influence factor data corresponding to each subarea in a target area, wherein the types of the influence factor data at least comprise daily maximum temperature, daily minimum temperature, single month accumulated high Wen Tianshu, single month accumulated low temperature days, town rate, resident population, economic acceleration and industrial electricity utilization structure;
according to an empirical mode decomposition technology, processing each historical power load data to obtain corresponding component sequence data;
determining main influence factor data corresponding to each component sequence data based on the similarity between each influence factor data and each component sequence data;
constructing a corresponding component prediction model by using the main influence factor data and the component sequence data;
and determining the power load prediction result of the target area according to the component prediction results of all the component prediction models.
2. The load prediction method according to claim 1, wherein the processing each of the historical power load data according to the empirical mode decomposition technique to obtain corresponding component sequence data comprises:
adding Gaussian white noise for preset times to the power load data to be decomposed to form a plurality of subsequences to be decomposed;
when empirical mode decomposition is performed, performing empirical mode decomposition on the subsequence to be decomposed, taking the average value of the obtained first modal components as a first modal component sequence of power load data to be decomposed, and calculating to obtain a corresponding first residual signal;
after adding a first modal component of Gaussian white noise into the first residual signal, continuing empirical mode decomposition, taking the average value of the obtained second modal component as a second modal component sequence of power load data to be decomposed, and calculating to obtain a corresponding second residual signal;
repeating the empirical mode decomposition until the empirical mode decomposition stopping condition is satisfied.
3. The load prediction method according to claim 2, wherein the determining the main influence factor data corresponding to each of the component sequence data based on the similarity between each of the influence factor data and each of the component sequence data, specifically comprises:
sequencing the influence factor data based on the time axis sequence to obtain a corresponding influence factor sequence;
carrying out standardization treatment on the influence factor sequence;
respectively calculating the similarity between the component sequence data and the influence factor sequences after each normalization processing;
and determining the influence factor sequence corresponding to the similarity result being larger than a preset similarity threshold value as main influence factor data.
4. A load prediction method according to claim 3, wherein the calculating of the similarity between the component sequence data and the influence factor sequence after each normalization processing, respectively, specifically comprises:
the similarity between the component sequence and the influencing factor sequence is calculated according to the following cumulative distance formula:
γ(i,j)＝d(f i ,p j )+min{γ(i-1,j-1),γ(i-1,j),γ(i,j+1)}
wherein, gamma is the accumulated distance, f i For the ith normalized component value in the component sequence F, the component sequence f= { F 1 ,f 2 ,…,f i ,…,f n }，p j To influence the j-th normalized component value in the factor sequence P, the factor sequence p= { P is influenced 1 ,p 2 ,…,p j ,…,p m }，d(f i ,p j ) For the distance of each component in the sequence of components F and the sequence of influencing factors P from each other.
5. The load prediction method according to claim 4, wherein the constructing a corresponding component prediction model from the main influence factor data and the component sequence data specifically includes:
dividing a set formed by the component sequence data and the main influence factor data into a training set and a testing set;
and inputting the samples of the training set into a neural network model for training, taking the mean square error as a training index, selecting the parameters corresponding to the minimum mean square error to generate a component prediction model, and evaluating through the test set.
6. The load prediction method according to claim 5, wherein the evaluating by the test set specifically comprises:
the mean absolute error percentage and the mean square error calculated according to the following formula are evaluated:
wherein MAPE is the average absolute error percentage, MSE is the mean square error, n is the number of data, y i As the real data of the data to be processed,is predictive data.
7. The load prediction method according to claim 6, wherein the determining the power load prediction result of the target area according to the component prediction results of all the component prediction models specifically includes:
obtaining a component prediction result corresponding to each component prediction model;
and carrying out summation processing on all the component prediction results to obtain the power load prediction result of the target area.
8. A load predicting apparatus, comprising:
the data acquisition module is used for respectively acquiring historical power load data and influence factor data corresponding to each subarea in the target area, wherein the types of the influence factor data at least comprise daily maximum temperature, daily minimum temperature, single month accumulated high Wen Tianshu, single month accumulated low temperature days, town ratio, resident population, economic acceleration and industrial electricity utilization structure;
the decomposition module is used for respectively processing each historical power load data according to an empirical mode decomposition technology to obtain corresponding component sequence data;
the main influence factor module is used for determining main influence factor data corresponding to each component sequence data based on the similarity between each influence factor data and each component sequence data;
the model construction module is used for constructing a corresponding component prediction model by the main influence factor data and the component sequence data;
and the load prediction module is used for determining the power load prediction result of the target area according to the component prediction results of all the component prediction models.
9. A load predicting device comprising a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the load predicting method according to any one of claims 1 to 7 when executing the computer program.
10. A computer readable storage medium, wherein the computer readable storage medium stores a computer program, and wherein the load prediction method according to any one of claims 1 to 7 is implemented when the computer program is executed by a device in which the computer readable storage medium is located.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311443662.1A CN117543544A (en) | 2023-10-31 | 2023-10-31 | Load prediction method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311443662.1A CN117543544A (en) | 2023-10-31 | 2023-10-31 | Load prediction method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117543544A true CN117543544A (en) | 2024-02-09 |
Family
ID=89785244
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311443662.1A Pending CN117543544A (en) | 2023-10-31 | 2023-10-31 | Load prediction method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117543544A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117852928A (en) * | 2024-03-08 | 2024-04-09 | 国网北京市电力公司 | Near zero energy consumption building load prediction method, device, equipment and medium |
| CN117932345A (en) * | 2024-03-08 | 2024-04-26 | 深圳国瑞协创储能技术有限公司 | Power load data prediction model group training method, device, equipment and medium |
| CN117852928B (en) * | 2024-03-08 | 2024-06-04 | 国网北京市电力公司 | Near zero energy consumption building load prediction method, device, equipment and medium |
-
2023
- 2023-10-31 CN CN202311443662.1A patent/CN117543544A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117852928A (en) * | 2024-03-08 | 2024-04-09 | 国网北京市电力公司 | Near zero energy consumption building load prediction method, device, equipment and medium |
| CN117932345A (en) * | 2024-03-08 | 2024-04-26 | 深圳国瑞协创储能技术有限公司 | Power load data prediction model group training method, device, equipment and medium |
| CN117852928B (en) * | 2024-03-08 | 2024-06-04 | 国网北京市电力公司 | Near zero energy consumption building load prediction method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109492830B (en) | Mobile pollution source emission concentration prediction method based on time-space deep learning | |
| CN111861013B (en) | Power load prediction method and device | |
| CN111027772A (en) | Multi-factor short-term load prediction method based on PCA-DBILSTM | |
| CN111126704A (en) | Multi-region precipitation prediction model construction method based on multi-graph convolution and memory network | |
| CN112734135B (en) | Power load prediction method, intelligent terminal and computer readable storage medium | |
| CN110210656B (en) | Shared bicycle flow prediction method and system based on station behavior analysis | |
| CN113449919B (en) | Power consumption prediction method and system based on feature and trend perception | |
| CN111461445B (en) | Short-term wind speed prediction method and device, computer equipment and storage medium | |
| CN117543544A (en) | Load prediction method, device, equipment and storage medium | |
| CN113344288B (en) | Cascade hydropower station group water level prediction method and device and computer readable storage medium | |
| CN113033861A (en) | Water quality prediction method and system based on time series model | |
| Li et al. | A novel multichannel long short-term memory method with time series for soil temperature modeling | |
| CN112508299A (en) | Power load prediction method and device, terminal equipment and storage medium | |
| CN111985719A (en) | Power load prediction method based on improved long-term and short-term memory network | |
| CN116187835A (en) | Data-driven-based method and system for estimating theoretical line loss interval of transformer area | |
| CN116799796A (en) | Photovoltaic power generation power prediction method, device, equipment and medium | |
| CN113268929B (en) | Short-term load interval prediction method and device | |
| CN111141879A (en) | Deep learning air quality monitoring method, device and equipment | |
| CN112990530B (en) | Regional population quantity prediction method, regional population quantity prediction device, electronic equipment and storage medium | |
| CN117114190A (en) | River runoff prediction method and device based on mixed deep learning | |
| CN116665798A (en) | Air pollution trend early warning method and related device | |
| CN116542701A (en) | Carbon price prediction method and system based on CNN-LSTM combination model | |
| CN115423159A (en) | Photovoltaic power generation prediction method and device and terminal equipment | |
| CN114386332A (en) | Soil humidity prediction method and device based on EDC-LSTM model and storage medium | |
| WO2022222230A1 (en) | Indicator prediction method and apparatus based on machine learning, and device and storage medium |
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 |