Disclosure of Invention
In order to meet the development requirement of the prior art, the invention provides a method for filling power curve data of a power grid user.
The invention provides a method for filling power curve data of power grid users, which is improved in that the method comprises the following steps:
estimating the line loss rate according to the state of the line T of the transformer area, and estimating the missing user power according to the daily curve state of the user;
and fine-tuning the estimated line loss rate and the estimated missing user power.
Further, the estimating of the line loss rate includes:
sample R is calculated as followsdmLine loss rate r at time (d, m) ∈ Gdm0:
Wherein, the sample R
dm=[r
dm1,r
dm2,…,r
dmN],P
dmT: the total power of the platform area; r is
dmn: the ratio of the power of user n at time m of day d to the total power of the cell,
P
dmn: the power of user n; p
dm0: line loss power; n: the number of users under the cell; n: the user number is an integer from 1 to N; time (d, m): record at time m on day D, D ∈ D, D: a set of recording days; m is belonged to M, M: recording a set of moments; g: and the time instants of all user power records in the station area are aggregated.
Further, the estimating of the line loss rate includes:
time of day
Sample R
dmIf the data is not complete, the recorded data is obtained in u users by the K-nearest neighbor algorithm of' city block distance
The nearest k is selected from the middle section
1A sample
Sample R is estimated as follows
dmLine loss rate of
In the formula, i: days, i belongs to D; j: the time number, j belongs to M; n is
1,n
2,…,n
u: represents u users;
user n at ith and jth time
uThe ratio of the power of (a) to the total power of the cell;
line loss rate of u users at the j time of day i.
Further, the estimating of the missing user power comprises:
let the power curve record of the user n on day d be L
dn=[P
d1n,P
d2n,…,P
dMn]At the moment of time
Recording power day curve of user t on ith day through K-neighbor algorithm of' correlation distance
K with the largest correlation of medium section
2A sample
Estimating missing user power data as follows
In the formula, m
lThe time of data acquisition;
t user records m on day i
wPower at time point, w is 1,2,3 …, l; l: the number of samples;
further, before the fine tuning, estimating a sum of the missing powers according to the user power record and the estimated line loss rate as follows
Wherein, e: user n with recorded power
1,n
2,…,n
u(ii) a f: user q missing power recording
1,q
2,…,q
v;
With incompletely recorded time
While, the sample R
dmThe line loss rate of (a).
Further, the fine tuning includes:
at the moment of time
Selection of k
2An
As required to be padded
Is estimated value of
Respectively calculate
And are combined with
Comparing, and setting the difference value as minimum
Further, the line loss rate is calculated
Setting a threshold r
1,r
2And are combined with
And comparing and determining the final estimated line loss rate:
if it is
The estimated value of the line loss rate is
If it is
The estimated value of the line loss rate is
If it is
The estimated value of the line loss rate is
Further, based on the estimated line loss rate
The sum of the missing user powers is calculated as follows
For estimated user power
Scaling to obtain final missing power estimate
A data padding apparatus for a power grid consumer power curve, the apparatus comprising:
the estimating unit is used for estimating the line loss rate according to the state of the line T of the transformer area and estimating the missing user power data according to the state of the daily curve of the user;
and the adjusting unit is used for fine-tuning the estimation of the line loss rate and the estimation of the power of the missing user to complete the filling of the missing data.
Further, the estimation unit includes:
the first estimation subunit is used for estimating the line loss rate according to the state of the line T of the transformer area;
the second estimation subunit is used for estimating the missing user power data according to the user daily curve state;
the adjusting unit includes:
the first adjusting subunit is used for determining a final estimated value of the line loss rate according to the set threshold and the corrected line loss rate;
and the second adjusting subunit is used for calculating the final estimated value of the power value of the missing user according to the final estimated value of the line loss rate.
Compared with the closest prior art, the technical scheme provided by the invention has the following excellent effects:
(1) according to the technical scheme provided by the invention, the line loss can be estimated without classifying the station area states, and the station area time with the closest state can be quickly and conveniently found through a neighbor algorithm; the daily curves of the users do not need to be classified by electricity utilization behaviors, the most similar daily curves can be directly matched for missing power data filling, the accuracy and the effectiveness of data filling are effectively improved, and data support is provided for data analysis of the power system.
(2) According to the technical scheme provided by the invention, the state and the missing data are respectively estimated in a horizontal and vertical combination mode through two dimensional directions, and fine adjustment is comprehensively considered, so that a better and more stable result can be obtained, and the correctness of curve data filling can be effectively improved.
Detailed Description
The technical solution provided by the present invention will be described in detail by way of specific embodiments in conjunction with the accompanying drawings of the specification.
The technical scheme provided by the invention is used for filling data aiming at the problem of the power utilization curve data loss of users in the power grid region. In the process of collecting curve data such as current, power and the like of a power user, a missing phenomenon usually occurs, taking the power curve data as an example, the technical scheme provided by the invention comprises the following steps:
knowing a certain phase line T of a single-phase or multi-phase power grid area, N consumers C are connected below1,C2,…,CN. Intelligent electric meter every other certain time t0Minute recording power per user, M points per day, where M x t0Data for D days, i.e., M × D recording times, are recorded continuously at 1440. For the recording at time (d, m), i.e. at time m on day d, the output power of T is PdmTAre all known; the power of N users is Pdm1,Pdm2,…,PdmNIs fully known or partially known; line loss power of Pdm0Not less than 0 and satisfies Pdm0=PdmT-(Pdm1+Pdm2+…+PdmN) When the power of all users in the station area is known, the line loss can be obtained through calculation. The problem is that the user power data P of the missing part needs to be filled updmn。
The method for filling the power curve data loss of the power grid user is based on a K-nearest neighbor algorithm, and in the method, the total power P of the platform area is only subjected to the condition that the denominator is 0 to prevent the occurrence of the condition that the denominator is 0dmTRecording > 0 for subsequent work, for total power PdmTFor a record of 0, all the missing user powers are noted as 0 below. Firstly, estimating a line loss rate by estimating a state of a station area T, and then estimating missing user power data by estimating a state of a user daily curve, wherein a technical route diagram is shown as an attached figure 1, and the specific flow is as follows:
(1) estimating the line loss rate r according to the state of the line T in the transformer areadm0
For the power recording at the time (d, m), the ratio of the power of user n to the total power of the cell is
Estimating the line loss rate of the current moment by the ratio of each user power to the total power of the cell, wherein the specific details are as follows:
corresponding to the time (d, m)Is Rdm=[rdm1,rdm2,…,rdmN]Wherein N represents the number of users using electricity, and G is a time set with complete power records of all users in the cell, that is, for the time (d, m) belonging to G, the power of all users in the cell is known, so the sample data R is the sample data R at this timedmIs complete, the line loss rate can be calculated as follows:
will { (R)ij,rij0) L (i, j) e } as a known (sample, line loss rate) set, where i: days, i belongs to D; j: at time j ∈ M.
For the time of day
The power of the user under the cell is not recorded completely, so the sample R
dmIncomplete, line loss rate unknown, following for R
dmAnd estimating the line loss rate. Set a moment
U users n are recorded
1,n
2,…,n
uPower of (1), selecting R
dmIn which there is a recorded part
And the same sorting operation is performed for all known samples,
through "
Distance between city blocks"K-nearest neighbor method, on a sorted known sample
In search of the nearest k
1Individual section sample
Using their corresponding original copies
To the sample R of the arithmetic mean of the line loss rates of
dmIs preliminarily estimated, i.e.
(2) Estimating missing user power data from user daily curve state
Let L be the power daily curve record of day d of user ndn=[Pd1n,Pd2n,…,PdMn]Estimating the missing part power of the incomplete daily power curve according to the known complete daily power curve of the user, wherein the specific details are as follows:
for the moment when a record is incomplete
Estimating each missing user power record P separately
dmq. Due to P
dmqAbsence of (2), daily curve L
dqDefinitely incomplete, sorting L
dqThe recorded part of the method is as follows:
let H
dqIs L
itAll of
And P
imtAll have daily curve sets recorded and are sorted
Wherein m is
lThe time of data acquisition; p
imlj: t user records m on day i
lPower at a time point; l: the number of samples; get the known (sample, label) set { (L)
it,P
imt)|,t)∈H
dq}. Through "
Correlation Distance of sex"K-nearest neighbor method, as knownSample(s)
Finding the k with the largest correlation
2A sample
Estimating P by sample total power ratio
dmqK of (a)
2The number of the candidate values is determined,
(3) fine-tuning correction of padding data
Set a moment
U users n are recorded
1,n
2,…,n
uPower of, missing v users q
1,q
2…, qv. From the known user power record and the line loss rate estimated by (1), the sum of the missing powers can be estimated
Wherein, in the step (A),
missing user q
fThe power of (d);
line loss rate average of incomplete sample data. Obtaining a plurality of estimates based on the estimate of the missing power in (2)
j=1,2,…,v,u=1,2,…,k
2(ii) a Wherein the content of the first and second substances,
missing user q
fAn estimate of (d). The line loss rate estimation is taken as a main factor, the missing power estimation and the line loss rate estimation are finely adjusted to obtain the final line loss rate and the estimation of the user power data, and a technical diagram is shown in fig. 2, and the specific details are as follows:
at the moment of time
For each one requiring padding
Can select k
2An
One of them is used as an estimated value
In total k
2 vAnd (6) generating an estimated value. Respectively calculate
Heel
The comparison is carried out, and the difference is set to be the minimum
Calculating the line loss rate estimated therefrom
And
make a comparison and set 2 thresholds r
1,r
2:
If it is not
The end of the line loss rateEstimated as
If it is not
The final estimate of the line loss rate is modified
If it is not
The final estimate of the line loss rate is modified
According to the finally estimated line loss rate
Calculate the sum of the missing user powers:
then to the preliminarily estimated user power
Scaling to obtain a final missing power estimate:
and finally completing the filling work of all missing data.
A device for filling power curve data of power grid users comprises:
the estimating unit is used for estimating the line loss rate according to the state of the line T of the transformer area and estimating the missing user power data according to the state of the daily curve of the user;
the estimation unit includes:
the first estimation subunit is used for estimating the line loss rate according to the state of the line T of the transformer area;
the second estimation subunit is used for estimating the missing user power data according to the user daily curve state;
and the adjusting unit is used for fine-tuning the estimation of the line loss rate and the estimation of the power of the missing user to complete the filling of the missing data.
The adjusting unit includes:
the first adjusting subunit is used for determining a final estimated value of the line loss rate according to the set threshold and the corrected line loss rate;
and the second adjusting subunit is used for calculating the final estimated value of the power value of the missing user according to the final estimated value of the line loss rate.
Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art can make modifications and equivalents to the embodiments of the present invention without departing from the spirit and scope of the present invention, which is set forth in the claims of the present application.