CN113159427A - Regional power grid load prediction method - Google Patents
Regional power grid load prediction method Download PDFInfo
- Publication number
- CN113159427A CN113159427A CN202110450560.7A CN202110450560A CN113159427A CN 113159427 A CN113159427 A CN 113159427A CN 202110450560 A CN202110450560 A CN 202110450560A CN 113159427 A CN113159427 A CN 113159427A
- Authority
- CN
- China
- Prior art keywords
- model
- power grid
- regional power
- grid load
- gray
- 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 51
- 230000005611 electricity Effects 0.000 claims description 30
- 230000002087 whitening effect Effects 0.000 claims description 14
- 238000006467 substitution reaction Methods 0.000 claims description 13
- 238000004519 manufacturing process Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 8
- 230000004044 response Effects 0.000 claims description 7
- 230000000694 effects Effects 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000001556 precipitation Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 230000001364 causal effect Effects 0.000 description 11
- 238000012360 testing method Methods 0.000 description 10
- 239000000243 solution Substances 0.000 description 7
- 238000010248 power generation Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 230000006872 improvement Effects 0.000 description 5
- 230000007547 defect Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000000611 regression analysis Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000013112 stability test Methods 0.000 description 2
- 239000007983 Tris buffer Substances 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005312 nonlinear dynamic Methods 0.000 description 1
- XOFYZVNMUHMLCC-ZPOLXVRWSA-N prednisone Chemical compound O=C1C=C[C@]2(C)[C@H]3C(=O)C[C@](C)([C@@](CC4)(O)C(=O)CO)[C@@H]4[C@@H]3CCC2=C1 XOFYZVNMUHMLCC-ZPOLXVRWSA-N 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000002195 synergetic effect Effects 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- 238000000700 time series analysis 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
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A regional power grid load prediction method comprises the steps of collecting regional power grid load historical data, establishing a gray Verhulst model: collecting historical data of regional power grid load, and establishing a system dynamics model: and determining the weight coefficients of the gray Verhulst model and the system dynamics model according to the minimum variance criterion to obtain a combined model, and realizing regional power grid load prediction according to the combined model. On the basis of not increasing complexity, the invention linearly combines the single models through the minimum variance criterion to establish a combined model for regional power grid load prediction, and is suitable for load prediction with complex structure, rich original information and close relation among subsystems.
Description
Technical Field
The invention belongs to the field of power load prediction, and relates to a regional power grid load prediction method.
Background
Under the background of high growth in domestic economy, the demand of power consumption is rapidly increased, and the difference of social development levels in different regions can lead to different power consumption. The load prediction work which is used as the key for optimizing, dispatching and planning construction of the power grid is an important premise and foundation for planning, running, dispatching and operating the regional power grid, and has important significance for improving the stability and economy of the running of the power grid, reducing the construction cost of a power transmission line and realizing the balance of power supply and demand.
The load prediction method of the power system mainly comprises a regression analysis method, a time series analysis method, an artificial neural network method, an index smoothing method and the like. The regression analysis method comprises the steps of establishing a regression model by analyzing historical data of a load, predicting the load by using the regression model, wherein the prediction precision is often unsatisfactory because the regression model is influenced by uncertainty of influence factors; the time series method and the Kalman filtering algorithm are combined and applied to short-term load prediction, and a prediction result with higher precision is obtained; research on the application of the artificial neural network method in medium-long term load prediction has also made relevant progress. The current load prediction method mainly has the following defects:
(1) the model parameters are determined according to historical data and cannot take into account the future development condition of the region.
(2) Various factors influencing the social electricity consumption cannot be fully considered.
(3) The non-parametric model is ambiguous in its physical meaning, making it more difficult to analyze and adjust the model itself when the prediction bias is large.
Disclosure of Invention
The invention aims to provide a regional power grid load prediction method aiming at the defects that the single model prediction error is large and the historical data cannot be fully utilized in the traditional regional power grid load prediction, the gray Verhulst and system dynamics combined model is utilized, the load prediction data obtained by the gray Verhulst model and the system dynamics model can be coupled to the final load prediction result according to a linear combination method, and the prediction result of the combined model can not only fit the historical load data, but also consider the variation trend of power consumption and subsystems thereof. The weight coefficient of the single model is obtained through the minimum variance criterion, so that the variance of the predicted value of the combined model is not larger than that of any single model, and the prediction accuracy of the combined model is improved as much as possible.
In order to achieve the purpose, the technical scheme of the invention is as follows:
a regional power grid load prediction method comprises the following steps:
step 1: collecting regional power grid load historical data, and establishing a gray Verhulst model:
step 2: collecting historical data of regional power grid load, and establishing a system dynamics model:
and step 3: and determining the weight coefficients of the gray Verhulst model and the system dynamics model according to the minimum variance criterion to obtain a combined model, and realizing regional power grid load prediction according to the combined model.
The invention has the further improvement that the specific process of the step 1 is as follows: acquiring regional power grid load historical data, processing the regional power grid load historical data to obtain an accumulated new sequence and a mean sequence, obtaining a gray differential equation according to the accumulated new sequence and the mean sequence, combining the gray differential equation with a whitening differential equation, solving to obtain a discrete solution of the whitening differential equation, and obtaining a load prediction time sequence, thereby obtaining a gray Verhulst model.
The further improvement of the invention is that the gray differential equation of the gray Verhulst model is as follows:
x0(k)+az1(k)=b(z1(k))2,k=2,3,...n (6)
in the formula: x is the number of0(k) Is the kth element in the regional power grid load historical data sequence; a. b is an unknown parameter; z is a radical of1(k) Is the kth element in the next-to-average generation sequence;
z1(k)=αx1(k)+(1-α)x1(k-1),k=2,3,...,n (5)
in the formula: α is a generation coefficient, x1(k) Is the k-th element in the new sequence is accumulated once;
in the formula: x is the number of0(i) Is the ith element in the regional power grid load historical data sequence.
A further improvement of the invention is that the whitening differential equation is:
in the formula: x is the number of1(t) is the time response; a. b is an unknown parameter.
The invention has the further improvement that the specific process of the step 2 is as follows: collecting and processing the historical data of the regional power grid load, screening power consumption influence factors, carrying out measurement economics analysis to obtain a power consumption equation, and obtaining a system dynamics model of power consumption according to the power consumption equation.
A further improvement of the invention is that the electricity consumption equation is:
in the formula: eDGenerating capacity for distributed energy;is a function that returns the natural logarithm of x; eREThe energy source is an energy source substitution effect; eRNIs the rate of re-electrification; eULThe utilization rate of fossil energy is improved; e1、E2、E3、E4Respectively being first industry power consumption, second industry power consumption, third industry power consumption and life power consumption; e2’、E3’、E4' annual electricity consumption of second industry, third industry and domestic electricity consumption respectively; g1、G2、G3Respectively the initial domestic production total value of the first industry, the second industry and the third industry of the economic body; r is the precipitation; g is the initial domestic production total value of the economic body; p is the number of the human mouths; t isavIs the average annual temperature.
The invention is further improved in that in step 3, the combined model predicted value is
Ecomb=w1E1+w2E2 (18)
In the formula: ecombIs a predicted value of the combined model; w is a1、w2Weight coefficients of a gray Verhulst model and a system dynamics model are respectively obtained; e1、E2Predicted values for the gray Verhulst model and the system dynamics model, respectively.
The invention is further improved in that the weight coefficients of the gray Verhulst model are as follows:
in the formula: var () is a variance function; cov () is a covariance function; w is a1Weight coefficients of a gray Verhulst model; r is1、r2The prediction residuals for the gray Verhulst model and the system dynamics model, respectively.
The invention is further improved in that the weight coefficients of the gray Verhulst model are as follows:
in the formula: var () is a variance function; cov () is a covariance function; w is a1Weight coefficients of a gray Verhulst model; r is1、r2The prediction residuals for the gray Verhulst model and the system dynamics model, respectively.
Compared with the prior art, the invention has the following beneficial effects: the gray prediction system established by the invention has the advantages of less original information required by prediction, simple calculation process and verifiable prediction result, and is suitable for prediction of load increase according to an S-shaped curve or load increase in a saturation stage. The invention analyzes the adaptability of different models, the combined model comprehensively utilizes all data of a single model on the basis of not increasing the complexity, the model information is enriched, the weight coefficient of the single model is determined by the minimum variance criterion, the variance of the residual error of the predicted value of the combined model is not larger than that of the residual error of the predicted value of any single model, the obtained load prediction curve is smoother, the risk that the single model is easy to generate larger prediction error is reduced, and the prediction precision is improved as much as possible.
Furthermore, the system dynamics model established by the invention can comprehensively consider the influence of economy, population, energy substitution, electrification and the like on the social electricity consumption, and is suitable for load prediction with complex structure, rich original information and close connection among subsystems.
Drawings
FIG. 1 is an integration curve of the Verhulst model.
FIG. 2 is a topological structure diagram of system dynamics.
FIG. 3 is a combined model of regional grid load prediction.
Fig. 4 is a flow chart of regional power grid load prediction based on a combined model.
Detailed Description
The invention is further described in detail below with reference to the figures and specific examples.
Referring to fig. 4, the process of the load prediction method for the regional power grid provided by the invention is as follows: the method comprises the steps that historical social electricity consumption data are used as original data, a gray differential equation and a whitening differential equation are sequentially established and solved, and a load prediction time sequence based on a gray Verhulst model is obtained, wherein the gray Verhulst model is suitable for prediction of load increase according to an S-shaped curve or load increase in a saturation stage; the influence of economy, population, energy substitution, electrification and the like on social electricity consumption is comprehensively considered, an economic subsystem, a population subsystem, an energy substitution and electrification subsystem and an electric power consumption subsystem for load prediction are established, and a load prediction model based on system dynamics is obtained and is suitable for load prediction with a complex structure, rich original information and close connection among subsystems. On the basis of not increasing complexity, linear combination is carried out on the single model through a minimum variance criterion, and a combined model for regional power grid load prediction is established. The method specifically comprises the following steps:
step I: establishing a gray Verhulst model
Collecting regional power grid load historical data, processing the regional power grid load historical data to obtain an accumulated new sequence and a mean value sequence, obtaining a gray differential equation of a gray Verhulst model, obtaining a whitening differential equation of the gray Verhulst model, solving to obtain a discrete solution of the whitening differential equation, finally obtaining a load prediction time sequence, and establishing the gray Verhulst model.
Step II: establishing a system dynamics model
And collecting historical data of the regional power grid load, processing the historical data of the regional power grid load, screening power consumption influence factors, performing measurement economics analysis to obtain power consumption and a subsystem equation thereof, and finally obtaining a system dynamics model of the power consumption.
Step III: establishing a combined model
Determining weight coefficients of a gray Verhulst model and a system dynamics model according to a minimum variance criterion to obtain a power consumption equation of the combined model, and obtaining each power generation and utilization, a single model predicted value and a combined model predicted value by using simulation software Vensim PLE.
Specifically, the method comprises the following steps:
step I: establishment of gray Verhulst model
(1) Introduction of the Verhulst model:
Pierrea common sigmoidal curve, the Logistic function, was named by Verhulst when studying population changes during the years 1838-1847, as follows:
in the formula: p (t) is population size; l is the maximum of the curve; k is the growth rate of the curve; t is t0Is the initial time; t is time. PierreVerhulst adds a natural resource limitation population development term in an exponential model of population growth dp (t)/dt ═ kp (t), so that the population growth rate k shows a linear decrease with the increase of the population number p (t), namely k ═ a ═ kp (t)1-A2X (t), thereby obtaining a Verhulst model:
in the formula: p (t) is population size; a. the1Is a linear parameter; a. the2Is a linear parameter; t is time. The formula (2) is a first-order autonomous differential equation, and an analytic solution P (t) can be obtained by a separation variable method:
in the formula: p (t) is population size; a. the1Is a linear parameter; a. the2Is a linear parameter; p0Is an initial value of population; t is time. To visually observe the population quantity over timeFrom a geometric point of view, the MATLAB drawing scheme (2) is used (the parameters in the scheme are as follows: A)1=1,A2=3,t0=0,P00.1) to obtain a geometric solution as shown in fig. 1.
(2) Establishment of a gray Verhulst model:
modeling of the gray Verhulst model requires that raw data must be equally spaced in time, and the processing idea is to accumulate the raw data (discrete historical data x0(1), x0(2), …, x0(n) required by the gray Verhulst model), weaken random factors of the raw time series data, and then establish a differential equation for generating numbers. Let the known sequence be x0(1),x0(2),…,x0(n), accumulating once to generate new sequence x1(1),x1(2),…,x1(n) wherein
In the formula: x is the number of1(k) Is the k-th element in the new sequence is accumulated once; x is the number of0(i) Is the ith element in the original known sequence. The original known sequence is regional power grid load historical data.
From the sequence x1(k) N is 1,2, …, and generates a sequence of close-proximity mean generation
z1(k)=αx1(k)+(1-α)x1(k-1),k=2,3,...,n (5)
In the formula: z is a radical of1(k) Is the kth element in the next-to-average generation sequence; α is a formation coefficient, 0 ≦ α ≦ 1, and generally, α may be 0.5; x is the number of1(k) Is to accumulate the kth element in the new sequence once. Establishing a gray differential equation
x0(k)+az1(k)=b(z1(k))2,k=2,3,...n (6)
In the formula: x is the number of0(k) Is the kth element in the regional power grid load historical data sequence; a. b is an unknown parameter; z is a radical of1(k) Is next to the kth element in the mean-generating sequence. The whitening differential equation for the corresponding Verhulst model is:
in the formula: x is the number of1(t) is the time response; a. b is an unknown parameter.
The grey differential equation is used to obtain parameters a and b, which are then introduced into the whitening differential equation to obtain the time response x1(t) arranging the formula (6) in a matrix form
Writing equation (8) as
Ax=η (9)
In the formula: a is a coefficient matrix containing an adjacent mean generation sequence; x is a column vector containing unknown parameters a, b; η is the column vector containing the original known sequence.
In general, rank (A) ≠ rank (A, eta), and equation (9) is a system of incompatible equations, so that only a least squares solution can be obtained, definedIs a least squares solution of equation (9). When this time, the formula (9) is changed to
In the formula: a is a coefficient matrix containing an adjacent mean generation sequence; a. theTIs the transpose of matrix a;is comprised of least squares parametersA column vector of (a); η is the column vector containing the original known sequence.
Least square solution toSubstituting the obtained signal into a whitening differential equation to obtain
Solving equation (11) yields a discrete time response sequence
Performing once subtraction reduction on the formula (12) to obtain a predicted sequence of the gray Verhulst model original sequence:
the gray Verhulst model is actually a method for finding out the change relation of a system from one or more discrete arrays of the system by using a number, is a model for trying to establish continuous change of the system, mainly reflects the processes of generation, development and saturation of things, and is used for predicting an S-shaped saturation load curve.
Step II: establishment of system dynamics model
(1) Brief introduction of system dynamics:
the system dynamics is developed by integrating system theory, control theory, servo mechanics, information theory, decision theory and computer simulation in 1950 by the american college of science and engineering, and the research on the system can be divided into the following 2 steps.
a. Dividing the system S into n mutually associated subsystems S according to a correlation theoryi。
b. Subsystem SiAnd modeling. The subsystem consists of a basic loop and a feedback loop, and the variables mainly comprise flow, volume, rate and auxiliary variables.
The understanding of the system dynamics to the problem is obtained through the process of establishing and manipulating a mathematical model based on the mutual close dependence relationship between the system behavior and the internal mechanism, and the causal relationship generating the change form is gradually discovered, and the system dynamics is called as the structure. The structure refers to a network formed by a set of action or decision rules that are linked by loops.
(2) Establishment of system dynamics model
The invention selects the following power consumption influence factors: total domestic production value (GDP), population quantity, population structure, electricity utilization for three-generation, electricity utilization for life, electrification process, energy substitution effect and the like. And stability test, synergy test and causal test are adopted for the power consumption influence factors. Through the cluster analysis of the power consumption influence factors, the regional power grid load prediction system can be divided into 4 subsystems: an economy subsystem, a population subsystem, an energy replacement and re-electrification subsystem and an electric power consumption subsystem, and a topological structure relationship among the subsystems is shown in figure 2.
a. An economic subsystem. The economic subsystem is derived from a model of the Soro growth with technological advances.
The economic subsystem equation is designed as:
in the formula: delta I is the investment increment; i isRThe investment rate is; a. theDThe asset depreciation rate; a is an asset; delta E is employment increment; Δ L is labor population increment; l isRThe labor increase rate; Δ W is the payroll increase;is a function used to obtain a time average and represent a desire; wGIs the payroll growth rate;is a function that returns the natural logarithm of x; wBIs the initial value of the payroll; delta TRIs the technology growth rate;is a function of the third order exponential delay of the return input;a function that returns the power of x of the base; t is a technical initial value;<T’>predicting the time length for the set load, wherein the unit is set as year; Δ GiThe total value increment of domestic production in the ith industry; giThe total value of domestic production of the ith industry is the initial value of the economic body; delta G is the total value increment of domestic production; g is the initial domestic production total value of the economic body;is a function that returns the accumulated sum of array indices from 1 to i.
b. A population subsystem. The influence of changes in population number and structure on GDP and power consumption is simulated based on an exogenous population growth model. The population quantity and the urbanization process influence the population of each product and employment, so that the population subsystem is connected with the economic subsystem, and the population quantity influences the domestic electricity consumption, so that the population subsystem is connected with the electricity consumption subsystem.
Population subsystems are derived from exponential growth models with natural resource constraints. The Logistic population growth model is a nonlinear dynamic model established for the total population PWherein n is the natural growth rate of the population, and A is the number of stable population allowed by the environmental resources. The equation for the change of population with time isP0Is the number of people at the initial time. The model generally reflects the natural birth of the population andenvironmental resource restrictions result in death. The natural growth rate of the population is determined by the model of the growth of the lolo, i.e. exogenous. The population mechanical growth rate refers to the change of population quantity caused by population migration in and out in a certain period of time, the population mechanical growth is generally represented by migration of economic laggard areas to economic developed areas, and the GDP is divided by the total population number to obtain the per-capita GDP influence factor, so that the population mechanical growth rate of the areas is influenced. The domestic electricity will be affected by the population and the GDP and the invention will be explained in detail in the electricity consumption subsystem.
The population subsystem equation is designed as:
in the formula: mGRMechanical growth rate of population;is a function of the integral of the return ratio; p is the number of the human mouths; g is the initial domestic production total value of the economic body; i isAVIs a human-averaged GDP influencing factor; n is a radical ofGRIs the natural growth rate of the population;is a function used to obtain a time average and represent a desire; n is a radical ofINIInitial value of natural population growth rate;<T’>predicting the time length for the set load, wherein the unit is set as year; u shapeGRThe urbanization rate;is a function that returns x to the power e; p2、P3The employment population numbers of the second industry and the third industry are respectively; pINIThe population number at the current time is an initial value; eLThe power is used for life;is a function that returns the natural logarithm of x;a function of a first order exponential delay for the return input; t isAVIs the average annual temperature.
c. Energy replacement and re-electrification subsystems. The energy substitution and re-electrification subsystem mainly simulates the access of new energy at a power generation side and the wide use of electric energy at a power utilization side, and is connected with the electric power consumption subsystem through the generated energy of distributed energy. The energy substitution and re-electrification subsystem equation is designed as
In the formula: eREThe energy source is an energy source substitution effect; ePOInfluence of energy policy;is a function that returns the natural logarithm of x; pV、HY、WPRespectively the growth rates of photovoltaic power generation, hydroelectric power generation and wind power generation; eCLGenerating capacity for clean energy;is a function of the integral of the return ratio; eCEDThe rate of increase of the demand for clean energy; eFEGenerating power for fossil energy; eULThe utilization rate of fossil energy is improved; eFEDIncreasing the fossil energy demand; eRNIs the rate of re-electrification;is a function of the power of x that returns the base.
d. An electrical power consuming subsystem. The electric power consumption subsystem mainly simulates the influence of the electricity consumption of the three-generation, the domestic electricity and the distributed energy generation amount on the final electric power consumption. The synergy and causal relationships between gdp (g), electricity consumption (E), assets (a), employment population (L) were analyzed using a method of economies of measure.
Equations (14), (15) and (16) relate variables of each subsystem, and the system dynamics simulation software is substituted with the equations, so that the variation of each subsystem can be simulated according to the equations. The economic subsystem, the population subsystem, the energy replacement and re-electrification subsystem and the later-mentioned electricity consumption subsystem are connected through variables, and the connection of the four subsystems is shown in detail in fig. 2.
The present invention will study the synergistic relationship between power consumption and economic growth and analyze whether power consumption causes economic growth, vice versa, or both, by causal tests. The co-integration of gdp (g), electricity consumption (E), assets (a), employment population (L) data was examined based on the methodology in the economies of measure, and the results of the JJ co-integration examination are shown in table 1, which shows that at least one set of co-integration exists at the 5% level.
TABLE 1 results of the collaborative alignment test by JJ method
After the coordination check, the causal relationship and direction between the variables also need to be judged. Firstly, the stationarity of time series variables is checked, unit root checks are carried out on GDP (G), power consumption (E), assets (A) and employment population (L), and the results show that the series are first-order and single-order. The causal relationships between gdp (g), electricity consumption (E), assets (K), employment population (L) were examined based on methods in the economies of measure, and Granger's causal test results are shown in table 2, which show that electricity consumption (E) and gdp (g), assets (a) have bilateral Granger causal relationships at a 10% confidence level, but no causal relationship from gdp (g) to electricity consumption (E).
TABLE 2 results using Granger causal test
Since there are too many factors affecting power consumption, if all the factors are considered in the system dynamics model, not only the model becomes complicated, but also the accuracy may decrease with distance from the current year, the main influencing factors are found out by using the correlation test method, and the power consumption equation is obtained from the test result, which is shown in table 3.
TABLE 3 correlation of Power consumption with its influencing factors
From the test results of table 3, the power consumption equation can be obtained:
in the formula: eDGenerating capacity for distributed energy;is a function that returns the natural logarithm of x; eREThe energy source is an energy source substitution effect; eRNIs the rate of re-electrification; eULThe utilization rate of fossil energy is high. E1、E2、E3、E4Respectively being first industry power consumption, second industry power consumption, third industry power consumption and life power consumption; e2’、E3’、E4' annual electricity consumption of second industry, third industry and domestic electricity consumption respectively; g1、G2、G3Respectively the initial domestic production total value of the first industry, the second industry and the third industry of the economic body; r is the precipitation; g is the initial domestic production total value of the economic body; p is the number of the human mouths; t isavIs the average annual temperature.
Step III: building of combined model
(1) Design of combined model
Substituting the equation of a gray Verhulst model and the equation of a system dynamics model into simulation software Vensim PLE to respectively obtain results E of the two prediction methods1And E2The prediction residuals are r1And r2. The combined prediction method obtains the predicted value of the combined model by solving the weighted arithmetic mean of the predicted value of the single model, the combined model provided by the invention simultaneously adopts the system dynamics and the gray Verhulst model for prediction, and then the two prediction results are linearly combined by adopting the minimum variance criterion to obtain the predicted value E of the combined modelcomb. The prediction residual is r, and the weight coefficients of the single model are respectively taken as w1And w2Satisfy 0<r1<1,0<r2<1,r1+r21. The combined model has a predicted value of
Ecomb=w1E1+w2E2 (18)
In the formula: ecombIs a predicted value of the combined model; w is a1、w2Weight coefficients of a gray Verhulst model and a system dynamics model are respectively obtained; e1、E2Predicted values for the gray Verhulst model and the system dynamics model, respectively.
Residual error of combined model is
r=w1r1+w1r2 (19)
In the formula: r is the prediction residual of the combined model; w is a1、w2Weight coefficients of a gray Verhulst model and a system dynamics model are respectively obtained; r is1、r2The prediction residuals for the gray Verhulst model and the system dynamics model, respectively.
Variance of residual error is
In the formula: var () is a variance function; cov () is a covariance function(ii) a r is the prediction residual of the combined model; w is a1、w2Weight coefficients of a gray Verhulst model and a system dynamics model are respectively obtained; r is1、r2The prediction residuals for the gray Verhulst model and the system dynamics model, respectively.
Wherein cov (r)1,r2) Is a residual error r1And r2The covariance of (a). A var (r) to w1Minimum value is calculated to obtain
In the formula: var () is a variance function; cov () is a covariance function; w is a1Weight coefficients of a gray Verhulst model; r is1、r2The prediction residuals for the gray Verhulst model and the system dynamics model, respectively.
It is obvious that the covariance between the single models can be taken to be 0 and has w1+w2When the weight coefficients of the combined model are 1, the weight coefficients are
In the formula: var () is a variance function; w is a1、w2Weight coefficients of a gray Verhulst model and a system dynamics model are respectively obtained; r is1、r2The prediction residuals for the gray Verhulst model and the system dynamics model, respectively.
The combined model equation is brought into simulation software VensimPLE, so that the prediction results of electricity for three generations, electricity for life and loads of different land blocks can be obtained, and the combined model is shown in figure 3.
(2) Use of composite models
And processing load historical data to obtain an accumulated new sequence and an adjacent mean value generation sequence, solving a gray differential equation to obtain a least square parameter, substituting the least square parameter into a whitening differential equation, solving to obtain a discrete time response sequence of load prediction, and performing first-order subtraction reduction to obtain a predicted value of a gray Verhulst model. When the economic subsystem and the population subsystem are processed, the equations obtained by the Solo growth model and the Logistic population growth model are substituted into the system dynamics model, when the power consumption subsystem is processed, stability test, coordination test and causal test are carried out on power consumption and influence factors thereof, and the power consumption equation is obtained through EViews software and substituted into the system dynamics model. And obtaining a weight coefficient of the single model by using a minimum variance criterion to obtain a power consumption equation of the combined model, substituting the obtained equations of the gray Verhulst model and the system dynamics model into simulation software VensimP PLE, and obtaining each power generation and utilization, a single model predicted value and a combined model predicted value by using the simulation software VensimP PLE.
The method is used for overcoming the defects that the single model prediction error is large in the traditional regional power grid load prediction and the historical data cannot be fully utilized. Establishing a gray differential equation and a whitening differential equation in sequence by taking historical social electricity consumption data as original data and solving to obtain a load prediction time sequence based on a gray Verhulst model, wherein the model is suitable for predicting that a load is increased according to an S-shaped curve or the load is in a saturation stage; the influence of economy, population, energy substitution, electrification and the like on social electricity consumption is comprehensively considered, an economic subsystem, a population subsystem, an energy substitution and electrification subsystem and an electric power consumption subsystem for load prediction are established, and a load prediction model based on system dynamics is obtained and is suitable for load prediction with a complex structure, rich original information and close connection among subsystems. On the basis of not increasing complexity, the method linearly combines the single models through the minimum variance criterion to establish a combined model for regional power grid load prediction.
Aiming at the defects that the single model prediction error is large and the historical data cannot be fully utilized in the traditional regional power grid load prediction, the regional power grid load prediction method is provided, and has the following advantages:
(1) the gray prediction system has the advantages of less original information required by prediction, simple calculation process and verifiable prediction result, and is suitable for prediction of load increase according to an S-shaped curve or load increase in a saturation stage. .
(2) The system dynamics model can comprehensively consider the influence of economy, population, energy substitution, re-electrification and the like on social electricity consumption, and is suitable for load prediction with complex structure, rich original information and close connection among subsystems.
(3) The single models are linearly combined through the minimum variance criterion, a combined model is established, load historical data can be fully utilized, and the risk that the single models generate large errors is avoided.
Claims (9)
1. A regional power grid load prediction method is characterized by comprising the following steps:
step 1: collecting regional power grid load historical data, and establishing a gray Verhulst model:
step 2: collecting historical data of regional power grid load, and establishing a system dynamics model:
and step 3: and determining the weight coefficients of the gray Verhulst model and the system dynamics model according to the minimum variance criterion to obtain a combined model, and realizing regional power grid load prediction according to the combined model.
2. The regional power grid load prediction method according to claim 1, wherein the specific process of the step 1 is as follows: acquiring regional power grid load historical data, processing the regional power grid load historical data to obtain an accumulated new sequence and a mean sequence, obtaining a gray differential equation according to the accumulated new sequence and the mean sequence, combining the gray differential equation with a whitening differential equation, solving to obtain a discrete solution of the whitening differential equation, and obtaining a load prediction time sequence, thereby obtaining a gray Verhulst model.
3. The regional power grid load prediction method of claim 2, wherein a gray differential equation of a gray Verhulst model is as follows:
x0(k)+az1(k)=b(z1(k))2,k=2,3,...n (6)
in the formula: x is the number of0(k) Is a regional power grid load historical data sequenceThe kth element; a. b is an unknown parameter; z is a radical of1(k) Is the kth element in the next-to-average generation sequence;
z1(k)=αx1(k)+(1-α)x1(k-1),k=2,3,...,n (5)
in the formula: α is a generation coefficient, x1(k) Is the k-th element in the new sequence is accumulated once;
in the formula: x is the number of0(i) Is the ith element in the regional power grid load historical data sequence.
5. The regional power grid load prediction method according to claim 1, wherein the specific process of the step 2 is as follows: collecting and processing the historical data of the regional power grid load, screening power consumption influence factors, carrying out measurement economics analysis to obtain a power consumption equation, and obtaining a system dynamics model of power consumption according to the power consumption equation.
6. The method for predicting the load of the regional power grid according to claim 5, wherein the power consumption equation is as follows:
in the formula: eDGenerating capacity for distributed energy;is a function that returns the natural logarithm of x; eREThe energy source is an energy source substitution effect; eRNIs the rate of re-electrification; eULThe utilization rate of fossil energy is improved; e1、E2、E3、E4Respectively being first industry power consumption, second industry power consumption, third industry power consumption and life power consumption; e2’、E3’、E4' annual electricity consumption of second industry, third industry and domestic electricity consumption respectively; g1、G2、G3Respectively the initial domestic production total value of the first industry, the second industry and the third industry of the economic body; r is the precipitation; g is the initial domestic production total value of the economic body; p is the number of the human mouths; t isavIs the average annual temperature.
7. The method for predicting the load of the regional power grid according to claim 1, wherein in the step 3, the prediction value of the combined model is
Ecomb=w1E1+w2E2 (18)
In the formula: ecombIs a predicted value of the combined model; w is a1、w2Weight coefficients of a gray Verhulst model and a system dynamics model are respectively obtained; e1、E2Predicted values for the gray Verhulst model and the system dynamics model, respectively.
8. The regional power grid load prediction method of claim 7, wherein the weight coefficients of the gray Verhulst model are as follows:
in the formula: var () is a variance function; cov () is a covariance function; w is a1As grey VerhulstA weight coefficient of the model; r is1、r2The prediction residuals for the gray Verhulst model and the system dynamics model, respectively.
9. The regional power grid load prediction method of claim 7, wherein the weight coefficients of the gray Verhulst model are as follows:
in the formula: var () is a variance function; cov () is a covariance function; w is a1Weight coefficients of a gray Verhulst model; r is1、r2The prediction residuals for the gray Verhulst model and the system dynamics model, respectively.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110450560.7A CN113159427A (en) | 2021-04-25 | 2021-04-25 | Regional power grid load prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110450560.7A CN113159427A (en) | 2021-04-25 | 2021-04-25 | Regional power grid load prediction method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113159427A true CN113159427A (en) | 2021-07-23 |
Family
ID=76870604
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110450560.7A Pending CN113159427A (en) | 2021-04-25 | 2021-04-25 | Regional power grid load prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113159427A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113780660A (en) * | 2021-09-13 | 2021-12-10 | 广东电网有限责任公司 | Resident electricity consumption prediction method, system and storage medium |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105678403A (en) * | 2015-12-30 | 2016-06-15 | 国网浙江省电力公司经济技术研究院 | Region saturation load prediction method based on model family decomposition and integration technology |
-
2021
- 2021-04-25 CN CN202110450560.7A patent/CN113159427A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105678403A (en) * | 2015-12-30 | 2016-06-15 | 国网浙江省电力公司经济技术研究院 | Region saturation load prediction method based on model family decomposition and integration technology |
Non-Patent Citations (1)
Title |
---|
李杰等: "地区电网负荷预测的灰色Verhulst与 系统动力学组合模型", 《西安交通大学学报》, vol. 55, no. 01, pages 68 - 77 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113780660A (en) * | 2021-09-13 | 2021-12-10 | 广东电网有限责任公司 | Resident electricity consumption prediction method, system and storage medium |
CN113780660B (en) * | 2021-09-13 | 2024-02-13 | 广东电网有限责任公司 | Residential electricity consumption prediction method, system and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108876054B (en) | Short-term power load prediction method based on improved genetic algorithm optimization extreme learning machine | |
CN112990556A (en) | User power consumption prediction method based on Prophet-LSTM model | |
CN112465664B (en) | AVC intelligent control method based on artificial neural network and deep reinforcement learning | |
Enke et al. | A hybrid neuro-fuzzy model to forecast inflation | |
CN111582534A (en) | Hydroelectric power generation prediction method based on extreme learning machine | |
CN112990500B (en) | Transformer area line loss analysis method and system based on improved weighted gray correlation analysis | |
CN111460001B (en) | Power distribution network theoretical line loss rate evaluation method and system | |
CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
CN111680786A (en) | Time sequence prediction method based on improved weight gating unit | |
CN113722997A (en) | New well dynamic yield prediction method based on static oil and gas field data | |
CN104834975A (en) | Power network load factor prediction method based on intelligent algorithm optimization combination | |
CN115759380A (en) | Short-term prediction method for inertia of power system based on SHAP-XGboost algorithm | |
CN114595861A (en) | MSTL (modeling, transformation, simulation and maintenance) and LSTM (least Square TM) model-based medium-and-long-term power load prediction method | |
CN113300380A (en) | Load curve segmentation-based power distribution network reactive power optimization compensation method | |
CN115640889A (en) | Power load prediction method based on multiple linear regression and improved LSTM | |
CN113159427A (en) | Regional power grid load prediction method | |
Wei et al. | A combination forecasting method of grey neural network based on genetic algorithm | |
Qu et al. | A trend forecast of import and export trade total volume based on LSTM | |
CN108615091B (en) | Electric power meteorological load data prediction method based on cluster screening and neural network | |
CN115759418A (en) | Adjustable capacity prediction method and system for electric power retail customer | |
CN114519460A (en) | Electric power prediction system based on discrete multivariable grey prediction | |
CN114266593A (en) | Power consumption prediction method based on KPCA and linear regression | |
Master et al. | An Empirical Study of the Relationship between Capital Structure and Financial Performance--Based on Neural Network Analysis | |
CN111626465A (en) | New energy power short-term interval prediction method and system | |
Sheng et al. | Research on Application of Neural Network Based on Cloud Computing in Power Load Forecasting |
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 |