CN109886495A - Electric grid investment prediction technique based on gray model - Google Patents
Electric grid investment prediction technique based on gray model Download PDFInfo
- Publication number
- CN109886495A CN109886495A CN201910147193.6A CN201910147193A CN109886495A CN 109886495 A CN109886495 A CN 109886495A CN 201910147193 A CN201910147193 A CN 201910147193A CN 109886495 A CN109886495 A CN 109886495A
- Authority
- CN
- China
- Prior art keywords
- electric grid
- grid investment
- time point
- value
- prediction
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a kind of electric grid investment prediction techniques based on grey Markov model, electric grid investment Specifications are chosen according to the degree of association first, obtain the numerical value of several time points corresponding electric grid investment volume and electric grid investment Specifications, it is constructed to obtain the gray model of electric grid investment prediction according to these historical datas, obtains the electric grid investment predicted value of predicted time point by Grey Model further according to the value of the electric grid investment Specifications of predicted time point.The present invention improves electric grid investment prediction accuracy, can also predict that value correction method is modified electric grid investment predicted value by the electric grid investment based on Markov model, further increase accuracy by improving to gray model.
Description
Technical field
The invention belongs to electric grid investment electric powder predictions, more specifically, are related to a kind of electricity based on gray model
Net investment forecasting method.
Background technique
Recently as the fast development of society, the continuous promotion of entire society's electricity consumption, so that power grid construction and investment
Scale is increasing.As soon as electric grid investment measuring and calculating be an important practical application engineering, how to improve the precision of investment evaluation with
And how to make electric grid investment strategy more scientific, it does not obtain also fine to solve at present.There are Multiple factors to act on shadow simultaneously
Ring the capital scale of power grid construction investment project to be measured, the scale of investment and influence electricity of each power grid project to be measured
Netting has uncertainty between each factor of corporate investment's measuring and calculating, and historical data relevant to electric grid investment measuring and calculating equally compares
Compared with shortage.
On the select permeability of prediction model, on the one hand traditional mathematical analysis method requires data volume big, on the other hand
It is required that it is independently of each other between in a linear relationship and each factor between each factor and investment, and in power grid asset investment forecasting field
In, it is not between stringent linear correlation and index between index of correlation and investment there is also interacting, and not only
It is vertical.It is irregular simultaneously for small data quantity and achievement data therefore, it is necessary to the deficiency that one kind both can overcome the disadvantages that traditional mathematics method
The investment forecasting method that condition can also adapt to very well.
Due to lacking reasonable technological means, it is simple that current existing electric grid investment budget is also in the traditional dependence of comparison
Mathematical measure calculate, and have very big subjective factor, investment decision is not scientific and reasonable enough.Since Utilities Electric Co.'s history is former
Cause, part index number data time statistics is less, and data volume is little, is difficult to be analyzed with traditional mathematics method.Therefore one is needed
The historical data information of the method synthetic finite of relatively complete science, the power grid asset that applies to for keeping it more scientific and reasonable are invested
In measuring and calculating.
Summary of the invention
The electric grid investment prediction based on gray model that it is an object of the invention to overcome the deficiencies of the prior art and provide a kind of
Method improves the accuracy of electric grid investment prediction.
For achieving the above object, the present invention is based on the electric grid investment prediction technique of gray model the following steps are included:
S1: the degree of association of each electric power network technique index and electric grid investment volume when operation of power networks is calculated, by electric power network technique index
Descending arrangement is carried out according to degree of association size, N-1 electric power network technique index is as electric grid investment Specifications before selecting;
S2: the numerical value of M time point corresponding electric grid investment volume and N-1 electric grid investment Specifications, note are obtained
Electric grid investment volume vector is X1=(x1(1),x1(2),…,x1(M)), wherein x1(j) electric grid investment at j-th of time point is indicated
Volume, j=1,2 ..., M, N-1 electric grid investment Specifications vector of note is respectively Xi=(xi(1),xi(2),…,xi
(M)), wherein xi(j) numerical value of (i-1)-th electric grid investment Specifications when j-th of time point, i=2,3 ..., N are indicated;
S3: the gray model of following electric grid investment prediction is constructed:
Wherein, e indicates that natural constant, a indicate development coefficient, biIndicate that cooperation index, c indicate initial value, use is following
Method is calculated:
By development coefficient a and N-1 cooperation index bnConstitute parameter column P=[a, b2,b3,…,bN]T, using following formula
The value of parameter column P is calculated:
P=(BTB)-1BTY
S4: the primary tired of the corresponding electric grid investment Specifications of predicted time point k and previous time point k-1 is obtained
Value added xi (1)(k) and xi (1)(k-1), the gray model in step S3 is substituted into, the electric grid investment predicted value of time point k and k-1 is obtained
One-accumulate valueThe then electric grid investment predicted value of time point k
The present invention is based on the electric grid investment prediction techniques of grey Markov model, choose power grid according to the degree of association first and throw
Specifications are provided, the numerical value of several time points corresponding electric grid investment volume and electric grid investment Specifications, root are obtained
It constructs to obtain the gray model of electric grid investment prediction according to these historical datas, further according to the electric grid investment correlation skill of predicted time point
Art refers to that target value obtains the electric grid investment predicted value of predicted time point by Grey Model.The present invention passes through to gray model
It improves, improves electric grid investment prediction accuracy, can also be repaired by the electric grid investment predicted value based on Markov model
Correction method is modified electric grid investment predicted value, further increases accuracy.
Detailed description of the invention
Fig. 1 is the specific embodiment flow chart of the electric grid investment prediction technique the present invention is based on gray model;
Fig. 2 is the flow chart of the electric grid investment prediction value correction method in the present embodiment based on Markov model;
Fig. 3 is the comparison of Grey Model value of the present invention in the present embodiment, traditional Grey Model value and true value
Curve graph.
Specific embodiment
A specific embodiment of the invention is described with reference to the accompanying drawing, preferably so as to those skilled in the art
Understand the present invention.Requiring particular attention is that in the following description, when known function and the detailed description of design perhaps
When can desalinate main contents of the invention, these descriptions will be ignored herein.
Embodiment
Fig. 1 is the specific embodiment flow chart of the electric grid investment prediction technique the present invention is based on gray model.Such as Fig. 1 institute
Show, the present invention is based on the electric grid investment prediction technique of gray model specifically includes the following steps:
S101: electric grid investment Specifications are determined:
The degree of association of each electric power network technique index and electric grid investment volume when calculating operation of power networks, by electric power network technique index according to
Degree of association size carries out descending arrangement, and N-1 electric power network technique index is as electric grid investment Specifications before selecting.N-1's
Value is determine according to actual needs.
It can according to need selection degree of association type in calculating correlation, select the Deng Shi degree of association, Deng in the present embodiment
Family name's degree of association is a kind of common degree of association, and details are not described herein for specific calculating process
S102: history electric grid investment data are obtained:
The numerical value of M time point corresponding electric grid investment volume and N-1 electric grid investment Specifications is obtained, remembers power grid
Investment vector is X1=(x1(1),x1(2),…,x1(M)), wherein x1(j) the electric grid investment volume at j-th of time point, j=are indicated
1,2 ..., M, N-1 electric grid investment Specifications vector of note is respectively Xi=(xi(1),xi(2),…,xi(M)), wherein xi
(j) numerical value of (i-1)-th electric grid investment Specifications when j-th of time point, i=2,3 ..., N are indicated.
S103: the gray model optimized based on square law is established:
In the present invention, by electric grid investment volume vector X1=(x1(1),x1(2),…,x1(M)) be considered as gray model is
System characteristic sequence, also referred to as behavior variable, each electric grid investment Specifications vector Xi=(xi(1),xi(2),…,xi(M))
It is higher because of prime sequences, also referred to as factor variable with system features serial correlation.Remember Xi (1)=(xi (1)(1),xi (1)(2),…,
xi (1)It (M)) is electric grid investment Specifications vector XiOne-accumulate sequence, xi (1)(j) it indicates the when j-th of time point
The one-accumulate value of i-1 electric grid investment Specifications, Z1 (1)=(z1 (1)(1),z1 (1)(2),…,z1 (1)It (M)) is power grid
Investment vector X1Close to average generation sequence, z1 (1)(j) indicate j-th time point electric grid investment volume close to mean value, then this
The gray model GM (1, N) of invention electric grid investment prediction is defined as:
Wherein, k indicates time point, x1(k) the electric grid investment volume at k-th of time point, z are indicated1 (1)(k) k-th of time is indicated
Point electric grid investment volume close to mean value, xi (1)(k) the one of k-th of time point, (i-1)-th electric grid investment Specifications are indicated
Secondary accumulated value, a indicate development coefficient, biIndicate cooperation index, P=[a, b2,b3,…,bN]TFor parameter column.
According to the estimated value of the available parameter column P of history electric grid investment data of step S102 are as follows:
P=(BTB)-1BTY (2)
Wherein:
The then approximate response of gray model GM (1, N) are as follows:
Wherein,Indicate k-th of time point corresponding electric grid investment volume one-accumulate value, e indicates natural constant.
It is rightGM (1, N) prediction model can be reduced by carrying out a regressive, i.e., the power grid at k-th time point is thrown
Money volume predicted valueExpression formula it is as follows:
Analysis mode (5), archetype assumesIt has passed through initial point (1, x1 (1)(1)), actual electric grid investment
Prediction model may might not pass through this point, therefore in order to improve the accuracy of electric grid investment prediction, the present invention uses mould
Analog values and the smallest principle of true value error sum of squares, optimize this value, the specific method is as follows:
Enable initial value c=x1 (1)(1),Then have:
To above formula derivation, and enable f'(c)=0 can obtain
The historical data substitution of step S102 can be obtained:
By the above-mentioned c that acquires instead of the x in formula (5)1 (1)(1) gray model optimized:
S104: electric grid investment prediction:
Obtain the one-accumulate of the corresponding electric grid investment Specifications of predicted time point k and previous time point k-1
Value xi (1)(k) and xi (1)(k-1), the gray model in step S3 is substituted into, the electric grid investment predicted value of time point k and k-1 is obtained
One-accumulate valueThe then electric grid investment predicted value of time point k
In order to which the value for predicting electric grid investment is more accurate, the present embodiment also proposed a kind of based on Markov model
Electric grid investment predicts value correction method.Fig. 2 is the electric grid investment predicted value amendment side in the present embodiment based on Markov model
The flow chart of method.As shown in Fig. 2, the electric grid investment in the present embodiment based on Markov model predicts the specific of value correction method
Step includes:
S201: reference data is obtained:
It is Q that reference data time points, which are arranged, and Q time is chosen in the time point of existing true electric grid investment value
Point.In general, the Q time point in order to keep prediction more accurate, preferred distance predicted time point k nearest.Remember this Q time
The true electric grid investment value of point is yq, wherein q=1,2 ..., Q, are corresponded to using the Grey Model that step S103 is obtained
Electric grid investment predicted value
S202: prediction error is calculated:
It calculates the corresponding electric grid investment of various time points and predicts error
S203: state interval is divided:
The minimum value and maximum value for remembering electric grid investment prediction error are respectively emin、emax, by prediction error intervals [emin,
emax] it is divided into H state interval Eh=[E1h,E2h], wherein h=1,2 ..., H.In order to realize prediction, the value of H should be greater than Q
The difference of range prediction time point k nearest time point k ' and predicted time point k, i.e. H > k-k ' in a time point.It is general next
It says, when state interval divides, each state interval is average.
S204: state transition probability matrix is obtained:
The state transfer case for predicting error between two time points is counted, is remembered by state EjState E is transferred to by h stepj′
Number be denoted as αjj′(h), wherein j, j '=1,2 ..., H, by state EjThe number for carrying out transition is denoted as βj, then predict error by
State EjState E is transferred to by h stepj′Probability be pjj′(h)=αjj′(h)/βj, obtain H state transition probability matrix P
(h):
S205: the amendment of electric grid investment predicted value:
The time point for being less than H with the difference of predicted time point k is selected from Q time point, remembers the quantity at these time points
For G, time point is denoted as kg, g=1,2 ..., G.According to the prediction error state and H state transition probability at this G time point
Matrix P (h) is obtained by this G time point kgPass through k-k when to predicted time point kgStep is transferred to the probability of each state interval,
The total probability for acquiring each state interval takes the corresponding state interval of maximum total probabilityPower grid as predicted time point k is thrown
The prediction error state section for providing predicted value, is calculated state intervalThe median for predicting error is e*, then according to following
Formula predicts the modified electric grid investment of electric grid investment predicted valueIt is modified, obtains revised electric grid investment prediction
Value
Technical effect in order to better illustrate the present invention verifies the present invention using a specific embodiment.This
With the electric grid investment data instance of certain Utilities Electric Co. 2005-2012 in embodiment.Table 1 is electric grid investment data in the present embodiment
Table.
Table 1
Data in table 1 are brought into formula (2) and the parameter of gray model is calculated is classified as: p=[1.9449,
7.3415,-0.5237,1.5757,-0.0572]T, substitute into formula (9) and acquire parameter c=8.7730.
Then 2005-2012 each year electric grid investment volume is predicted using gray model.In order to compare,
It is compared herein using the predicted value of traditional gray model without square law optimization.Fig. 3 is of the invention in the present embodiment
The contrast curve chart of Grey Model value, traditional Grey Model value and true value.Table 2 is present invention ash in the present embodiment
The predicted value error contrast table of color model and traditional gray model.
Table 2
According to Fig. 3 and table 2 it is found that gray model is after square law optimizes, prediction error has larger improvement.
Next the predicted value of gray model of the present invention is repaired using the prediction value correction method based on Markov
Just.Table 3 is power grid asset investment and Grey Model Modelling predicted value situation table.
Serial number | True value | Predicted value | Relative error/% | Markov state |
1 | 8.98 | 8.98 | 0 | 1 |
2 | 9.49 | 9.4553 | -0.731 | 1 |
3 | 10.34 | 10.6082 | 5.1868 | 3 |
4 | 12.43 | 12.2406 | -3.0472 | 1 |
5 | 12.67 | 12.7194 | 0.7802 | 2 |
6 | 14.36 | 14.5281 | 2.3416 | 2 |
7 | 15.96 | 15.8973 | -0.7862 | 1 |
Table 3
As shown in table 3, relative error is up to 5.1868, minimum -3.0472, therefore the opposite of Markov model is missed
Poor range is [- 3.0472,5.1868], and Markov state is divided into three sections accordingly.Table 4 is the Ma Er of the present embodiment
Section husband state demarcation table.
State | State boundaries |
E1 | (- 3.04, -0.31] |
E2 | (- 0.31, -2.45] |
E3 | (- 2.45,5.19] |
Table 4
According to the state demarcation section of 3 data of table and table 4, obtaining a step state transition probability matrix isObtaining 2,3 step state transition probability matrixs by a step state transition probability matrix is respectively
Table 5 be predict 2012 belonging to state transfer matrix table.
Table 5
It is E according to 5 2012 years most possible affiliated states of table1, amendment error is e*=(- 3.04%-
0.31%)/2=-1.675%, Grey Model value in 2012 is 17.0337, and the relative error with true value is
6.18%, final revised predicted value isIt is opposite with true value to miss
Difference is 4.582%.Traditional Grey Model value is 17.0326, and the relative error with true value is 6.21%.As it can be seen that by
After electric grid investment prediction value correction method based on Markov model is modified, the error of electric grid investment predicted value is further
Reduce.
Although the illustrative specific embodiment of the present invention is described above, in order to the technology of the art
Personnel understand the present invention, it should be apparent that the present invention is not limited to the range of specific embodiment, to the common skill of the art
For art personnel, if various change the attached claims limit and determine the spirit and scope of the present invention in, these
Variation is it will be apparent that all utilize the innovation and creation of present inventive concept in the column of protection.
Claims (3)
1. a kind of electric grid investment prediction technique based on gray model, which comprises the following steps:
S1: the degree of association of each electric power network technique index and electric grid investment volume when calculating operation of power networks, by electric power network technique index according to
Degree of association size carries out descending arrangement, and N-1 electric power network technique index is as electric grid investment Specifications before selecting;
S2: obtaining the numerical value of M time point corresponding electric grid investment volume and N-1 electric grid investment Specifications, remembers power grid
Investment vector is X1=(x1(1),x1(2),…,x1(M)), wherein x1(j) the electric grid investment volume at j-th of time point, j=are indicated
1,2 ..., M, N-1 electric grid investment Specifications vector of note is respectively Xi=(xi(1),xi(2),…,xi(M)), wherein xi
(j) numerical value of (i-1)-th electric grid investment Specifications when j-th of time point, i=2,3 ..., N are indicated;
S3: the gray model of following electric grid investment prediction is constructed:
Wherein, e indicates that natural constant, a indicate development coefficient, biIndicate cooperation index, c indicates initial value, using following methods meter
It obtains:
By development coefficient a and N-1 cooperation index bnConstitute parameter column P=[a, b2,b3,…,bN]T, it is calculated using the following equation
Obtain the value of parameter column P:
P=(BTB)-1BTY
S4: the one-accumulate value of the corresponding electric grid investment Specifications of predicted time point k and previous time point k-1 is obtained
xi (1)(k) and xi (1)(k-1), the gray model in step S3 is substituted into, the one of the electric grid investment predicted value of time point k and k-1 is obtained
Secondary accumulated valueThe then electric grid investment predicted value of time point k
2. electric grid investment prediction technique according to claim 1, which is characterized in that the electric grid investment that the step S4 is obtained
Predicted valueIt is modified using the electric grid investment prediction value correction method based on Markov model, the specific method is as follows:
S4.1: it is Q that the setting reference data time, which counts, and Q time is chosen in the time point of existing true electric grid investment value
Point;Remember that this Q time point true electric grid investment value is yq, wherein q=1,2 ..., Q, the gray model obtained using step S3
Prediction obtains corresponding electric grid investment predicted value
S4.2: it calculates the corresponding electric grid investment of various time points and predicts error
S4.3: the minimum value and maximum value of note electric grid investment prediction error are respectively emin、emax, by prediction error intervals [emin,
emax] it is divided into H state interval Eh=[E1h,E2h], wherein h=1,2 ..., H;
S4.4: predicting the state transfer case of error between two time points of statistics, remembers by state EjState E is transferred to by h stepj′
Number be denoted as αjj' (h), wherein j, j '=1,2 ..., H, by state EjThe number for carrying out transition is denoted as βj, then predict error by
State EjState E is transferred to by h stepj′Probability be pjj′(h)=αjj′(h)/βj, obtain H state transition probability matrix P
(h):
S4.5: the time point for being less than H with the difference of predicted time point k is selected from Q time point, remembers the quantity at these time points
For G, time point is denoted as kg, g=1,2 ..., G.According to the prediction error state and H state transition probability at this G time point
Matrix P (h) is obtained by this G time point kgPass through k-k when to predicted time point kgStep is transferred to the probability of each state interval,
The total probability for acquiring each state interval takes the corresponding state interval of maximum total probabilityPower grid as predicted time point k is thrown
The prediction error state section for providing predicted value, is calculated state intervalThe median for predicting error is e*, then according to following
Formula predicts the modified electric grid investment of electric grid investment predicted valueIt is modified, obtains revised electric grid investment prediction
Value
3. electric grid investment prediction technique according to claim 2, which is characterized in that Q time point in the step S4.1
For Q nearest time point of range prediction time point k.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910147193.6A CN109886495A (en) | 2019-02-27 | 2019-02-27 | Electric grid investment prediction technique based on gray model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910147193.6A CN109886495A (en) | 2019-02-27 | 2019-02-27 | Electric grid investment prediction technique based on gray model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109886495A true CN109886495A (en) | 2019-06-14 |
Family
ID=66929731
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910147193.6A Pending CN109886495A (en) | 2019-02-27 | 2019-02-27 | Electric grid investment prediction technique based on gray model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109886495A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111598475A (en) * | 2020-05-22 | 2020-08-28 | 浙江工业大学 | Power grid risk prediction method based on improved gray Markov model |
CN113592176A (en) * | 2021-07-29 | 2021-11-02 | 国网新疆电力有限公司经济技术研究院 | Power grid production technical improvement project investment prediction method |
-
2019
- 2019-02-27 CN CN201910147193.6A patent/CN109886495A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111598475A (en) * | 2020-05-22 | 2020-08-28 | 浙江工业大学 | Power grid risk prediction method based on improved gray Markov model |
CN113592176A (en) * | 2021-07-29 | 2021-11-02 | 国网新疆电力有限公司经济技术研究院 | Power grid production technical improvement project investment prediction method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108230049A (en) | The Forecasting Methodology and system of order | |
CN108846517B (en) | Integration method for predicating quantile probabilistic short-term power load | |
CN110969290B (en) | Runoff probability prediction method and system based on deep learning | |
CN104504619B (en) | Two kinds consider that the monthly system of temperature and economic growth factor calls power predicating method | |
CN112507610B (en) | Interval prediction method for hot rolling coiling temperature | |
CN111353656A (en) | Steel enterprise oxygen load prediction method based on production plan | |
CN108876021B (en) | Medium-and-long-term runoff forecasting method and system | |
CN108537379B (en) | Self-adaptive variable weight combined load prediction method and device | |
CN106897774B (en) | Multiple soft measurement algorithm cluster modeling methods based on Monte Carlo cross validation | |
CN103942422B (en) | Granular-computation-based long-term prediction method for converter gas holder positions in metallurgy industry | |
CN102663224A (en) | Comentropy-based integrated prediction model of traffic flow | |
CN112149879A (en) | New energy medium-and-long-term electric quantity prediction method considering macroscopic volatility classification | |
CN109886495A (en) | Electric grid investment prediction technique based on gray model | |
CN106600959A (en) | Traffic congestion index-based prediction method | |
CN103294928A (en) | Combination forecasting method of carbon emission | |
CN111476677A (en) | Big data-based electricity consumption type electricity sales quantity analysis and prediction method and system | |
CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
CN104881707A (en) | Sintering energy consumption prediction method based on integrated model | |
CN109214709B (en) | Method for optimizing distribution of oxygen generation system of iron and steel enterprise | |
CN108734340A (en) | A kind of river flood forecasting procedure generally changed based on big vast type | |
CN107293118A (en) | A kind of traffic speed motion interval Forecasting Approach for Short-term | |
CN110020475A (en) | A kind of Markov particle filter method of forecasting traffic flow | |
CN104881718A (en) | Regional power business index constructing method based on multi-scale leading economic indicators | |
CN111639111A (en) | Water transfer engineering-oriented multi-source monitoring data deep mining and intelligent analysis method | |
CN114169434A (en) | Load prediction method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190614 |