CN113610288A - Power demand prediction method, device and storage medium - Google Patents
Power demand prediction method, device and storage medium Download PDFInfo
- Publication number
- CN113610288A CN113610288A CN202110855808.8A CN202110855808A CN113610288A CN 113610288 A CN113610288 A CN 113610288A CN 202110855808 A CN202110855808 A CN 202110855808A CN 113610288 A CN113610288 A CN 113610288A
- Authority
- CN
- China
- Prior art keywords
- power demand
- model
- data
- demand prediction
- energy
- 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 67
- 238000003860 storage Methods 0.000 title claims description 17
- 230000007613 environmental effect Effects 0.000 claims abstract description 17
- 238000010238 partial least squares regression Methods 0.000 claims description 26
- 238000003062 neural network model Methods 0.000 claims description 21
- 238000004519 manufacturing process Methods 0.000 claims description 17
- 238000005265 energy consumption Methods 0.000 claims description 12
- 238000007689 inspection Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000012847 principal component analysis method Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 238000007405 data analysis Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 2
- 238000009776 industrial production Methods 0.000 claims 3
- 230000009466 transformation Effects 0.000 abstract description 7
- 238000005215 recombination Methods 0.000 abstract description 2
- 230000006798 recombination Effects 0.000 abstract description 2
- 238000013528 artificial neural network Methods 0.000 description 23
- 238000012731 temporal analysis Methods 0.000 description 10
- 238000000700 time series analysis Methods 0.000 description 10
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000012417 linear regression Methods 0.000 description 6
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 238000013213 extrapolation Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 230000005611 electricity Effects 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 229910052799 carbon Inorganic materials 0.000 description 3
- 239000003345 natural gas Substances 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000000513 principal component analysis Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 1
- 241000255588 Tephritidae Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000001744 unit root test Methods 0.000 description 1
- 238000012795 verification Methods 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Pure & Applied Mathematics (AREA)
- Biophysics (AREA)
- Mathematical Optimization (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Computational Mathematics (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Evolutionary Biology (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Quality & Reliability (AREA)
- Water Supply & Treatment (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Probability & Statistics with Applications (AREA)
Abstract
The scheme discloses a power demand prediction method, which comprises the following steps: acquiring economic, social, energy and environmental data of a certain area, and establishing three single power demand prediction models; giving the power demand prediction model weight based on a shapey value method; and obtaining a power demand prediction equation after recombination, and predicting the power demand of a certain area based on the power demand prediction equation. The method provides a new combined model prediction method for power demand prediction, improves the accuracy of the prediction model, and considers the problem of energy transformation.
Description
Technical Field
The present invention relates to the field of power demand forecasting technologies, and in particular, to a power demand forecasting method, device and storage medium.
Background
The existing power prediction method can be divided into single model prediction and combined model prediction. Single model prediction is generally divided into conventional time series analysis methods and machine learning algorithms. The time series analysis method includes an elastic coefficient method, a differential autoregressive moving average model (ARIMA), an exponential smoothing method, a multiple linear regression method, and the like. The machine learning algorithm mainly comprises a BP neural network method, a gray prediction method, a support vector machine method, a wavelet analysis method and the like.
There are two types of combined predictive models: one is to optimize and modify a certain model, for example, a GA algorithm is used to modify a BP neural network model to form a GA-BP neural network model; and extracting data fluctuation characteristics by utilizing self-adaptive Fourier decomposition, and selecting a support vector machine for modeling by applying a sine and cosine optimization algorithm. The other type is that several models are endowed with different weights to be combined, for example, a new model is formed by recombination of an ARIMA model, a BP neural network model and a Holt exponential smoothing model; the method combines a Wavelet Neural Network (WNN) based on an improved IEMD model, an ARIMA model and an optimization algorithm of fruit flies into a new mixed model and the like.
At present, a single model prediction method is adopted for power demand prediction, and the prediction usually adopts a certain characteristic of data, so that various information of the data cannot be fully utilized, and the prediction precision is relatively low.
The combined predictive model ignores some types of each single model selected. For example, some methods only select a trend extrapolation method in time series analysis, and do not consider a linear regression method in time series analysis; the established electric power demand influence factor system only considers the conventional factors such as economy, society and weather, and does not consider the era background of energy transformation; when the model is verified, only data of a certain region is often adopted, and the problem of influence of region difference on the model applicability is not considered.
Disclosure of Invention
One purpose of the scheme is to provide a power demand prediction method, which provides a new combined model prediction method for power demand prediction, improves the accuracy of the prediction model, and considers the problem of energy transformation.
Another object of the present solution is to provide an apparatus for performing the above prediction method.
A third object of the present solution is to provide a storage medium.
In order to achieve the purpose, the scheme is as follows:
a method of power demand forecasting, the method comprising:
obtaining economic, social, energy and environmental data of a certain area, and establishing a plurality of single power demand prediction models;
giving the power demand prediction model weight based on a shapey value method;
and obtaining a power demand prediction equation based on the power demand prediction model weight, and predicting the power demand of a certain area based on the power demand prediction equation.
Preferably, the power demand prediction model comprises a partial least squares regression model, a differential auto-regression moving average model and a principal component BP neural network model.
Preferably, the establishing the partial least squares regression model comprises:
obtaining economic, social, energy and environmental data of a certain area;
performing grey correlation degree analysis on each item of acquired data to select key factors;
and selecting data with the gray correlation degree reaching a first preset value as an independent variable, constructing a partial least square regression equation, and establishing a partial least square regression model.
Preferably, the establishing the differential autoregressive moving average model comprises:
acquiring power demand data of a certain region in a certain period of time as original sample data;
performing time series data stationarity test and processing on the acquired original sample data to acquire stable time series data;
establishing a differential autoregressive moving average model based on the stable time sequence data, and training the differential autoregressive moving average model based on eviews 6.0;
and (5) carrying out model inspection.
Preferably, the establishing of the pivot BP neural network model includes:
obtaining economic, social, energy and environmental data of a certain area;
and analyzing the acquired data based on a principal component analysis method, and selecting principal components with interpretation capability reaching a second preset value as input layers of the BP neural network model to train the BP neural network model.
Preferably, the economic data of a certain region comprises a domestic total production value, a first industry production value ratio, a second industry production value ratio, a third industry production value ratio, a consumer price index and a production price index; the social data comprises population number, urbanization rate and road network density; the environmental data includes a summer maximum air temperature; the energy data comprises the gross domestic production energy consumption of ten thousand yuan and the primary energy consumption ratio of clean energy.
In a second aspect, there is provided an electric power demand prediction apparatus, comprising:
the data acquisition unit is used for acquiring economic, social, energy and environmental data of a certain region and establishing a plurality of power demand prediction models;
the data analysis unit is used for giving the power demand prediction model weight based on a shapey value method; and obtaining a power demand prediction equation based on the power demand prediction model weight, and predicting the power demand of a certain area based on the power demand prediction equation.
In a third aspect, a computer-readable storage medium is provided, on which a computer program is stored, which program, when being executed by a processor, is adapted to carry out the steps of the method according to any of the above.
The scheme has the following beneficial effects:
the power demand prediction model provides a new combined model prediction method for power demand prediction, and has high prediction precision. In the model selection, the model not only comprises a differential autoregressive moving average model (ARIMA) of a trend extrapolation method in a time series analysis method, a partial least squares regression model (PLS) of a multiple linear regression method, but also comprises a BP neural network model based on a machine learning algorithm. The three single predictive models include both "longitudinal" predictions using historical data and "lateral" predictions using various influencing factors.
Drawings
In order to illustrate the implementation of the solution more clearly, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the solution, and that other drawings may be derived from these drawings by a person skilled in the art without inventive effort.
FIG. 1 is a flow chart of a method for forecasting power demand;
FIG. 2 is a schematic diagram of a power demand prediction apparatus;
FIG. 3 is a flow chart of a prediction method in an embodiment;
FIG. 4 is a graph of auto-correlation coefficient DY and partial auto-correlation coefficient of a first-order differential term of power demand in a certain region according to an embodiment;
FIG. 5 is a plot of the white noise test of the residual term produced by the difference auto-regressive moving average model for a region according to an embodiment.
Detailed Description
Embodiments of the present solution will be described in further detail below with reference to the accompanying drawings. It is clear that the described embodiments are only a part of the embodiments of the present solution, and not an exhaustive list of all embodiments. It should be noted that, in the present embodiment, features of the embodiment and the embodiment may be combined with each other without conflict.
The terms "first," "second," and the like in the description and in the claims, and in the drawings described above, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Some types of the selected individual models are ignored due to the existing power demand prediction methods. For example, some methods only select the trend extrapolation method in the time series analysis, and do not consider the linear regression method in the time series analysis. The inventor comprehensively considers the selection of the influence factors and makes up the problem that the energy transformation is not considered in the existing research. A partial least squares regression (PLS) -differential autoregressive moving average (ARIMA) -principal element BP neural network combined model is established, a novel combined model prediction method is provided for power demand prediction, and the accuracy of the prediction model is improved.
As shown in fig. 1, a power demand prediction method includes the following steps:
s100, obtaining economic, social, energy and environmental data of a certain area, and establishing a plurality of single power demand prediction models including a partial least square regression model, a differential autoregressive moving average model and a principal element BP neural network model;
s200, endowing the power demand prediction model weight based on a shapey value method;
s300, obtaining a power demand prediction equation based on the power demand prediction model weight, and predicting the power demand of a certain area based on the power demand prediction equation.
Wherein establishing the partial least squares regression model comprises:
obtaining economic, social, energy and environmental data of a certain area;
performing grey correlation analysis on the acquired data;
and selecting data with the gray correlation degree reaching a first preset value as independent variables to construct a partial least square regression equation and establish a partial least square regression model.
Establishing the differential autoregressive moving average model comprises the following steps:
acquiring power demand data of a certain region in a certain period of time as sample data to establish a differential autoregressive moving average model;
the differential auto-regressive moving average model is trained based on eviews 6.0.
The establishing of the pivot BP neural network model comprises the following steps:
obtaining economic, social, energy and environmental data of a certain area;
and analyzing the acquired data based on a principal component analysis method, and selecting principal components with interpretation capability reaching a second preset value as input layers of the BP neural network model to train the BP neural network model.
The economic data of a certain area in the scheme comprises a domestic total production value, a first industry production value ratio, a second industry production value ratio, a third industry production value ratio, a consumer price index and a production price index; the social data comprises population number, urbanization rate and road network density; the environmental data includes a summer maximum air temperature; the energy data comprises ten thousand yuan of domestic total production energy consumption and the primary energy consumption ratio of clean energy.
As shown in fig. 2, an electric power demand prediction apparatus 1 includes:
the data acquisition unit 10 is used for acquiring economic, social, energy and environmental data of a certain region and establishing a plurality of power demand prediction models;
the data analysis unit 20 is configured to give the power demand prediction model weight based on a shapey value method; and obtaining a power demand prediction equation based on the power demand prediction model weight, and predicting the power demand of a certain area based on the power demand prediction equation.
On the basis of the above embodiment of the prediction method, in one embodiment, a computer-readable storage medium is further provided. The computer-readable storage medium is a program product for implementing the above-described identification method, which may employ a portable compact disc read only memory (CD-ROM) and include program code, and may be run on a device, such as a personal computer. However, the program product in this embodiment is not limited in this respect, and in this document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
A computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable signal medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as JAvA, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user computing device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).
The present solution will be described in detail with reference to fig. 3 to 5 and the obtained data of a certain region. The prediction is performed as the flow shown in fig. 3.
Firstly, a partial least squares regression (PLS) model is established
1. Obtaining various data of a certain area
Data of the four aspects of economy, society, energy and environment are obtained and used as influence factors on power demand.
The economic data includes:
total domestic product (GDP); the first industry output value accounts for the first industry, the second industry output value accounts for the second industry, and the third industry output value accounts for the third industry; consumer price index CPI; the production price index PPI is an index for measuring the change trend and change degree of the factory price of the products of the industrial enterprises.
The social data includes: the number of people; the urbanization rate; road network density;
the environmental data includes: the highest temperature in summer;
the energy data includes: the index refers to energy consumed by the gross domestic production of each ten thousand yuan; the proportion of the clean energy in the primary energy consumption is determined, the energy consumption is composed of the consumption of coal, petroleum, natural gas, primary electric power and the like, wherein the natural gas, the primary electric power and the like are clean energy, and the proportion of the natural gas, the primary electric power and the like in the total energy consumption can be used for measuring the proportion of the clean energy.
2. Analysis of Gray correlation
And calculating the grey correlation degree, selecting the influence factor with the grey correlation degree larger than 0.75 as a key influence factor, and taking the key influence factor as an independent variable of a partial least squares regression (PLS) model. And calculating the grey correlation degree of each influence factor by using the DPS data processing system. GDP (x1), a first industry output value ratio (x2), a second industry output value ratio (x3), a third industry output value ratio (x4), CPI (x5), PPI (x6), population number (x7), urbanization rate (x8), road network density (x9), summer maximum air temperature (x10) and unit GDP energy consumption (x11) are selected as key factors.
3. Constructing a partial least squares regression (PLS) equation, and establishing a partial least squares regression (PLS) model
Partial least squares regression analysis is carried out on key factors by using SIMCA-P software, a least squares regression equation shown as a formula (1) is established, and a standardized partial least squares model of the electricity demand and 11 influence factors in a certain area is obtained:
and (3) according to a normalized inverse process, reducing the partial least squares regression equation to obtain a formula (2):
4. model inspection
And (5) performing equation fitting effect analysis. And predicting the electricity consumption of a certain area in 2018 and 2019 by using the established partial least square regression equation, and calculating a relative error (the relative error is (predicted value-true value)/true value) to analyze whether the fitting degree of the regression equation is good or not. The relative error between the power demand predicted by the PLS model and the actual value is within 5%, which shows that the prediction precision of the model is high.
Secondly, establishing a differential autoregressive moving average (ARIMA) model 1, and acquiring data
The power demand of a certain area is acquired as sample data.
2. Stationarity test
The ARIMA model requires that the data sequence must be a stationary time sequence, so the data is firstly subjected to stationary inspection. Sequence stationarity was checked using ADF inspection. If the original sequence is not stable, the difference processing needs to be carried out on the original sample data until the original sample data is stable; and performing time series data stability inspection on the acquired original sample data, if the time series of the original sample data is not stable, performing first-order difference on the original sample data, if the time series of the first-order sample data obtained through the first-order difference is stable, using the first-order sample data as an independent variable, and if the time series of the first-order sample data obtained through the first-order difference is not stable, performing second-order difference on the original sample data, and inspecting again until the time series of the original sample data is stable.
The ADF unit root test is performed on the power demand, the ADF statistic is 1.460721, the ADF statistic is larger than the critical value under each significance level, the ADF statistic is an unstable time sequence, and the difference is needed to be performed on the original data. After differentiation, the ADF statistic was-4.7097, less than the threshold at each significance level, and the p value was 0.0005, which passed the ADF test. Therefore, d in ARIMA (p, d, p) is defined as 1.
3. Establishing an ARIMA model
And establishing a differential autoregressive moving average model based on the stable time sequence data, and training the differential autoregressive moving average model based on eviews 6.0.
And selecting the power demand in 1995 and 2017 as sample data to establish a model.
In the ARIMA model, the key is the order of p, q. The most common method is to use the tailing and truncation of the Autocorrelation Coefficient (ACF) and Partial Autocorrelation Coefficient (PACF) of the time series to judge the values of p and q. Specifically, the method comprises the following steps: when the autocorrelation coefficient is trailing, the partial autocorrelation coefficient is p-order truncated, and the model is AR (p); when the autocorrelation coefficient is a q-order truncation, the partial autocorrelation coefficient is trailing, and the model is MA (q); when the autocorrelation coefficients are q-order tails and the partial autocorrelation coefficients are p-order tails, the model is ARIMA (p, d, q). The autocorrelation coefficient and the partial autocorrelation coefficient of the first order differential term of the power demand of a certain region are shown in fig. 4:
fig. 4 shows that the autocorrelation coefficient and the partial autocorrelation coefficient of the first order differential term of the power demand in a certain area are both trailing, so an ARIMA (p,1, q) model should be established. Wherein, the partial autocorrelation coefficient gradually tends to 0 from 3-order lag, and the autocorrelation coefficient has a significant trend of tending to 0 after 3-order lag, so to avoid the influence of subjective factors, p is selected to [1,3], q is selected to [1,3], and a model is preliminarily set: ARIMA (1,1,1), ARIMA (1,1,2), ARIMA (1,1,3), ARIMA (2,1,1), ARIMA (2,1,2), ARIMA (2,1,3), ARIMA (3,1,1), ARIMA (3,1,2), ARIMA (3,1, 3). Firstly, as can be seen from p-value examination, at the significance level of 5%, the coefficients of the models except ARIMA (1,1,1), ARIMA (1,1,3), ARIMA (2,1,3) and ARIMA (3.1.3) are not significant and are removed first. The AIC and BIC criteria will be used herein to determine the optimal model, and generally, the smaller the AIC and SC statistics, the higher the accuracy of the corresponding model. Table 1 shows the AIC statistics and SC statistics for each model. As can be seen from table 1, ARIMA (1,1,1) has the smallest AIC statistic and SC statistic, so this embodiment selects ARIMA (1,1,1) as a model for predicting power demand in a certain area.
TABLE 1
When p is 1, d is 1, and q is 1, an equation shown in formula (3) may be established:
ΔYt＝41.5753+0.5896ΔYt-1+ut-0.9973ut-1 (3)
the formula (4) is obtained after simplification:
Yt＝41.5753+1.5896Yt-1-0.5896Yt-2+ut-0.9973ut-1 (4)
4. model inspection
One is the residual term white noise test. Theoretically, if the residual sequence is not a white noise sequence, it indicates that the information extracted by the model is incomplete, and the residual has some information and rules, and the model needs to be further modified. Fig. 5 shows that the autocorrelation coefficient and the partial autocorrelation coefficient of the residual sequence are both substantially within the confidence interval, so that it can be determined that the residual sequence is a white noise sequence, and the established model is good.
And secondly, analyzing the fitting effect of the equation. And predicting the power consumption of the three areas selected in 2018 and 2019 by using the established ARIMA model, and calculating a relative error (the relative error is (predicted value-true value)/true value) to analyze whether the fitting degree of the model is intact. Through calculation, the relative errors of the true value and the predicted value are within 5 percent, which indicates that the model is more accurate.
Thirdly, establishing a principal element BP neural network model
1. Principal component analysis
By using a principal component analysis method, principal component analysis is carried out on GDP (x1), a first industry output ratio (x2), a second industry output ratio (x3), a third industry output ratio (x4), CPI (x5), PPI (x6), population number (x7), urbanization rate (x8), road network density (x9), summer maximum air temperature (x10), unit GDP energy consumption (x11) and clean energy at primary energy ratio (x12), and a principal component with interpretation capability of 80% is selected. By utilizing eviews software, the interpretation capability of the first three main components y1, y2 and y3 reaches 90.54%, so that the main components y1, y2 and y3 are selected as input layers of the BP neural network.
2. Establishing principal element BP neural network model
The power demand and various influence factors in a certain area in the year 2000-2019 are selected as training samples and testing samples, and the matlab software is used for establishing, training and testing the BP neural network. Specifically, the method comprises the following steps: the principal components y1, y2 and y3 are used for predicting the power demand of a certain region, namely in a BP neural network, an input layer selects 3 neurons, and an output layer selects 1 neuron. According to the Kolmogorov theorem, the hidden layer selects 7 neural nodes. The remaining parameters are set as: the maximum number of training times is 1000 and the target error is 0.000001. The initial weight, learning speed, threshold value and the like of the network are automatically selected by Matlab software. And finally determining the weight from the input layer to the hidden layer, the threshold value of the hidden layer and the weight from the hidden layer to the output layer through training and continuous adjustment of the weight and the threshold value, and taking the weights as a model for the next test.
3. Model inspection
And analyzing the fitting effect of the model. And predicting the electricity consumption of the three areas in 2018 and 2019 by using the established pivot BP neural network model, and calculating a relative error (the relative error is (predicted value-true value)/true value) to analyze whether the fitting degree of the model is good or not. Through calculation, the relative errors of the true value and the predicted value are within 5 percent, which indicates that the model is more accurate.
Fourthly, a PLS-ARIMA-principal element BP neural network combined model is established
1. Calculating the weight of each model according to the shape value method
According to three single prediction models, calculating the power demand of a certain area in the year 2000-2017, comparing the power demand with a real value to obtain a relative error, and calculating by utilizing a mean square error (MAPE) criterion to obtain that the mean square errors of the PLS, the ARIMA and the principal element BP neural network are respectively as follows: 2.56%, 4.71%, 2.18%, with a total error of (2.56% + 4.71% + 2.18%)/3 ═ 3.15%. According to the calculation method of sharley value, the relative errors of all subsets can be obtained as shown in table 2:
TABLE 2
According to the method for calculating the sharey value, the sharey value of each of the three single prediction models can be obtained, and as shown in table 3, the weight of each of the three single prediction models can be obtained.
TABLE 3
The equation of the PLS-ARIMA-principal element BP neural network prediction model obtained finally is shown as a formula (5),
yt＝0.4y1t+0.15y2t+0.45y3t (5)
wherein y1t, y2t and y3t respectively refer to the predicted values of the 1 st, 2 nd and 3 rd prediction models at the time t.
2. Model inspection
And analyzing the fitting effect of the model. And predicting the electricity consumption of the three areas 2018 and 2019 by using the established combined prediction model, and calculating a relative error (the relative error is (predicted value-true value)/true value) to analyze whether the fitting degree of the model is good or not. Through calculation, the relative errors of the true value and the predicted value are within 5 percent, which indicates that the model is more accurate.
Compared with the existing combined prediction model, the PLS-ARIMA-principal element BP neural network combined model of the embodiment has the following advantages:
(1) existing power demand prediction methods ignore some types of single models that are selected. For example, some methods only select a trend extrapolation method in time series analysis, and do not consider a linear regression method in time series analysis;
the PLS-ARIMA-principal element BP neural network combined model provides a new combined model prediction method for power demand prediction, and has high prediction accuracy. In the model selection, the model not only comprises a differential autoregressive moving average model (ARIMA) of a trend extrapolation method in a time series analysis method, a partial least squares regression model (PLS) of a multiple linear regression method, but also comprises a BP neural network model based on a machine learning algorithm. The three single predictive models include both "longitudinal" predictions using historical data and "lateral" predictions using various influencing factors. The specific advantages of selecting the PLS model, the ARIMA model and the neural network model are as follows:
firstly, the power demand is influenced by a plurality of factors, and the PLS model not only can consider the influence factors of the power demand, but also can solve the problem of multiple collinearity among all variables; secondly, the power demand data is non-stationary sequence data, and the ARIMA model is a prediction model established aiming at the non-stationary sequence; finally, the principal component BP neural network can prevent the result of the output layer from being larger relative error caused by overhigh dimension of the input layer by analyzing the principal components of the variables of the input layer of the BP neural network.
(2) The existing electric power demand influence factor system only considers the conventional factors such as economy, society, weather and the like, and does not consider the era background of energy transformation. And the ten-thousand-yuan GDP energy consumption and the clean energy ratio representing low-carbon factors are incorporated into the paper, so that the background of energy transformation and accelerated development is reflected.
(3) When the existing model is verified, only data of a certain region are often adopted, and the problem of influence of region difference on model applicability is not considered.
In the model verification part, three regions with different regional characteristics are selected to verify the model, so that the persuasion of the model is enhanced.
(4) Although the use of neural networks is becoming widespread, the convergence problem has been one of the bottlenecks in its development. Convergence problems generally include two categories: the first is that neural networks are difficult to converge quickly; the second is that it is difficult for the neural network to converge to a globally optimal solution. The first category of problems is caused by a number of reasons, the most important of which is usually caused by training a user to input a huge amount of data into a neural network without screening. Therefore, the principal component analysis is introduced into the neural network, and the dimension of the input data of the neural network is reduced. The model built is called a principal element BP neural network.
The effect of the PLS-ARIMA-principal BP neural network and the conclusions drawn:
(1) the prediction precision of the PLS-ARIMA-principal element BP neural network is within 5%, which shows that the model can scientifically and accurately predict the power demand, is suitable for provinces with different regional characteristics, and provides a reliable prediction method for power prediction.
(2) By establishing a PLS-ARIMA-principal element BP neural network combined prediction model and comparing relative errors of the single prediction model and the combined prediction model, the prediction accuracy of the 3 single prediction models is verified to be not as high as that of the combined model, which shows that the combined prediction model not only can more fully utilize information provided by data, but also is obviously superior to the single prediction model in prediction accuracy.
(3) From the regression equation derived from the PLS model alone, it can be found that the main factors affecting the power demand are GDP, third-industry yield ratio and population size. Therefore, to accurately predict future power demand, the future trends of GDP, industry value ratio, and population size should be analyzed with emphasis. The specific gravity of the low carbon factor follows the specific gravity of the low carbon factor, and the influence of energy transformation is fully considered when power demand prediction is carried out in the future.
It should be understood that the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention, and it will be obvious to those skilled in the art that other variations or modifications may be made on the basis of the above description, and all embodiments may not be exhaustive, and all obvious variations or modifications may be included within the scope of the present invention.
Claims (8)
1. A method for predicting demand for electric power, the method comprising:
obtaining economic, social, energy and environmental data of a certain area, and establishing a plurality of power demand prediction models;
giving each power demand prediction model weight based on a shapey value method;
and obtaining a power demand prediction equation based on the weights of the power demand prediction models, and predicting the power demand of a certain area based on the power demand prediction equation.
2. The power demand prediction method of claim 1, wherein the power demand prediction model comprises a partial least squares regression model, a differential auto-regression moving average model, and a principal component BP neural network model.
3. The power demand prediction method of claim 2, wherein establishing the partial least squares regression model comprises:
obtaining economic, social, energy and environmental data of a certain area;
performing grey correlation degree analysis on each item of acquired data to select key factors;
and selecting data with the gray correlation degree reaching a first preset value as an independent variable, constructing a partial least square regression equation, and establishing a partial least square regression model.
4. The power demand prediction method of claim 2, wherein building the differential auto-regressive moving average model comprises:
acquiring power demand data of a certain region in a certain period of time as original sample data;
performing time series data stationarity test and processing on the acquired original sample data to acquire stable time series data;
establishing a differential autoregressive moving average model based on the stable time sequence data, and training the differential autoregressive moving average model based on eviews 6.0;
and (5) carrying out model inspection.
5. The power demand prediction method of claim 2, wherein building the principal component BP neural network model comprises:
obtaining economic, social, energy and environmental data of a certain area;
and analyzing the acquired data based on a principal component analysis method, and selecting principal components with interpretation capability reaching a second preset value as input layers of the BP neural network model to train the BP neural network model.
6. The power demand prediction method according to claim 1, wherein the economic data of the certain area includes a domestic total production value, a first industrial production value ratio, a second industrial production value ratio, a third industrial production value ratio, a consumer price index and a production price index; the social data comprises population number, urbanization rate and road network density; the environmental data includes a summer maximum air temperature; the energy data comprises the gross domestic production energy consumption of ten thousand yuan and the primary energy consumption ratio of clean energy.
7. An electric power demand prediction apparatus, comprising:
the data acquisition unit is used for acquiring economic, social, energy and environmental data of a certain region and establishing a plurality of power demand prediction models;
the data analysis unit is used for giving the plurality of power demand prediction model weights based on a shapey value method; and obtaining a power demand prediction equation based on the power demand prediction model weight, and predicting the power demand of a certain area based on the power demand prediction equation.
8. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110855808.8A CN113610288A (en) | 2021-07-28 | 2021-07-28 | Power demand prediction method, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110855808.8A CN113610288A (en) | 2021-07-28 | 2021-07-28 | Power demand prediction method, device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113610288A true CN113610288A (en) | 2021-11-05 |
Family
ID=78305779
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110855808.8A Pending CN113610288A (en) | 2021-07-28 | 2021-07-28 | Power demand prediction method, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113610288A (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103854073A (en) * | 2014-03-24 | 2014-06-11 | 昆明理工大学 | Method for comprehensively predicting generation capacity of multi-radial flow type small hydropower station group area |
| CN103853939A (en) * | 2014-03-24 | 2014-06-11 | 昆明理工大学 | Combined forecasting method for monthly load of power system based on social economic factor influence |
| CN106779157A (en) * | 2016-11-18 | 2017-05-31 | 中铁第勘察设计院集团有限公司 | The Forecasting Methodology of Regional Energy consumption demand |
| CN107590562A (en) * | 2017-09-05 | 2018-01-16 | 西安交通大学 | A kind of Short-Term Load Forecasting of Electric Power System based on changeable weight combination predicted method |
| CN107748938A (en) * | 2017-11-06 | 2018-03-02 | 国网福建省电力有限公司 | A kind of electric power demand forecasting method based on Vector Autoression Models |
| CN108320053A (en) * | 2018-01-23 | 2018-07-24 | 国网冀北电力有限公司经济技术研究院 | A kind of region electricity demand forecasting method, apparatus and system |
| CN110415013A (en) * | 2019-06-12 | 2019-11-05 | 河海大学 | A kind of combination forecasting method of net about vehicle trip requirements in short-term |
| CN110991739A (en) * | 2019-11-29 | 2020-04-10 | 南方电网能源发展研究院有限责任公司 | Construction method and prediction method of industry electric quantity demand prediction model |
| CN111401808A (en) * | 2020-03-12 | 2020-07-10 | 重庆文理学院 | Material agreement inventory demand prediction method based on hybrid model |
| CN112991088A (en) * | 2021-02-04 | 2021-06-18 | 国网浙江省电力有限公司 | Monthly power demand prediction method based on VMD-ANFIS-ARIMA |
-
2021
- 2021-07-28 CN CN202110855808.8A patent/CN113610288A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103854073A (en) * | 2014-03-24 | 2014-06-11 | 昆明理工大学 | Method for comprehensively predicting generation capacity of multi-radial flow type small hydropower station group area |
| CN103853939A (en) * | 2014-03-24 | 2014-06-11 | 昆明理工大学 | Combined forecasting method for monthly load of power system based on social economic factor influence |
| CN106779157A (en) * | 2016-11-18 | 2017-05-31 | 中铁第勘察设计院集团有限公司 | The Forecasting Methodology of Regional Energy consumption demand |
| CN107590562A (en) * | 2017-09-05 | 2018-01-16 | 西安交通大学 | A kind of Short-Term Load Forecasting of Electric Power System based on changeable weight combination predicted method |
| CN107748938A (en) * | 2017-11-06 | 2018-03-02 | 国网福建省电力有限公司 | A kind of electric power demand forecasting method based on Vector Autoression Models |
| CN108320053A (en) * | 2018-01-23 | 2018-07-24 | 国网冀北电力有限公司经济技术研究院 | A kind of region electricity demand forecasting method, apparatus and system |
| CN110415013A (en) * | 2019-06-12 | 2019-11-05 | 河海大学 | A kind of combination forecasting method of net about vehicle trip requirements in short-term |
| CN110991739A (en) * | 2019-11-29 | 2020-04-10 | 南方电网能源发展研究院有限责任公司 | Construction method and prediction method of industry electric quantity demand prediction model |
| CN111401808A (en) * | 2020-03-12 | 2020-07-10 | 重庆文理学院 | Material agreement inventory demand prediction method based on hybrid model |
| CN112991088A (en) * | 2021-02-04 | 2021-06-18 | 国网浙江省电力有限公司 | Monthly power demand prediction method based on VMD-ANFIS-ARIMA |
Non-Patent Citations (4)
| Title |
|---|
| 严国津: "基于Shapley值的中国电力需求组合预测模型", 《上海电力学院学报》, vol. 31, no. 6, pages 189 - 190 * |
| 杨实俊: "基于预测模型库关联优化的电力负荷组合预测模型", 《华北电力大学学报》, no. 1 * |
| 王涛: "基于Shapley组合模型的上海市电力需求趋势预测", 《智能计算机与应用》, vol. 9, no. 4, pages 33 - 39 * |
| 钟宏宇: "组合预测技术在风电中的研究分析", 《电器与能效管理技术》, no. 16 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110232203B (en) | Knowledge distillation optimization RNN short-term power failure prediction method, storage medium and equipment | |
| CN108830417B (en) | ARMA (autoregressive moving average) and regression analysis based life energy consumption prediction method and system | |
| CN106656357B (en) | Power frequency communication channel state evaluation system and method | |
| CN112116198A (en) | Data-driven process industrial state perception network key node screening method | |
| CN115423594A (en) | Enterprise financial risk assessment method, device, equipment and storage medium | |
| Salmanzadeh-Meydani et al. | The causal relationship among electricity consumption, economic growth and capital stock in Iran | |
| CN117132135A (en) | Urban informatization management system and method based on digital twinning | |
| CN116956189A (en) | Current abnormality detection system, method, electronic equipment and medium | |
| CN114819395A (en) | Industry medium and long term load prediction method based on long and short term memory neural network and support vector regression combination model | |
| CN116384843B (en) | Energy efficiency evaluation model training method and monitoring method for digital energy nitrogen station | |
| CN117725537A (en) | Real-time metering data processing platform | |
| Baur et al. | Explainability and interpretability in electric load forecasting using machine learning techniques–A review | |
| CN117520809A (en) | Transformer fault diagnosis method based on EEMD-KPCA-CNN-BiLSTM | |
| Chowhaan et al. | Machine learning approach for house price prediction | |
| CN116561569A (en) | Industrial power load identification method based on EO feature selection and AdaBoost algorithm | |
| CN113610288A (en) | Power demand prediction method, device and storage medium | |
| CN114970674A (en) | Time sequence data concept drift adaptation method based on relevance alignment | |
| CN113919599A (en) | Medium-and-long-term load prediction method | |
| Owda et al. | Using artificial neural network techniques for prediction of electric energy consumption | |
| He et al. | Probability density prediction of peak load based on mixed frequency noise-assisted multivariate empirical mode decomposition | |
| Amalnik et al. | Cash flow prediction using artificial neural network and GA-EDA optimization | |
| CN116780524B (en) | Industrial enterprise short-term load prediction method based on LSTM deep learning | |
| Vaghefzadeh et al. | Modelling customer demand for mobile value-added services: non-stationary time series models or neural networks time series analysis? | |
| Kurniasari et al. | Performance of the Accuracy of Forecasting the Consumer Price Index Using the Garch and Ann Methods | |
| Kyshenko et al. | Technological monitoring in the management of the distillation-rectification plant |
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 |