CN112036672B - New energy power generation ultra-short term power prediction method and system based on iterative correction - Google Patents

New energy power generation ultra-short term power prediction method and system based on iterative correction Download PDF

Info

Publication number
CN112036672B
CN112036672B CN202011227869.1A CN202011227869A CN112036672B CN 112036672 B CN112036672 B CN 112036672B CN 202011227869 A CN202011227869 A CN 202011227869A CN 112036672 B CN112036672 B CN 112036672B
Authority
CN
China
Prior art keywords
power
short
term
time
error
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.)
Active
Application number
CN202011227869.1A
Other languages
Chinese (zh)
Other versions
CN112036672A (en
Inventor
李登宣
程序
周海
崔方
吴骥
陈卫东
秦昊
居蓉蓉
朱想
丁煌
胡东平
秦放
丛从
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Electric Power Research Institute Co Ltd CEPRI
Original Assignee
China Electric Power Research Institute Co Ltd CEPRI
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China Electric Power Research Institute Co Ltd CEPRI filed Critical China Electric Power Research Institute Co Ltd CEPRI
Priority to CN202011227869.1A priority Critical patent/CN112036672B/en
Publication of CN112036672A publication Critical patent/CN112036672A/en
Application granted granted Critical
Publication of CN112036672B publication Critical patent/CN112036672B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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

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)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Water Supply & Treatment (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention discloses a new energy power generation ultra-short term power prediction method and system based on iterative correction, wherein the method comprises the following steps: acquiring absolute errors and relative errors of short-term predicted power and theoretical generated power at a predicted time and a plurality of times before the current time, and absolute errors and relative errors between the short-term predicted power and actual generated power; calculating a combined weighted absolute error and a combined weighted relative error based on the absolute error and the relative error between the short-term predicted power and theoretical generated power, the absolute error and the relative error between the short-term predicted power and actual generated power, and a weight coefficient obtained according to time sequence iteration; and correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error or the combined weighted relative error to obtain the short-term predicted power value of the predicted time. The ultra-short term prediction power value is predicted by a real-time online iterative correction method of the combined weighted error, and the ultra-short term power prediction precision of new energy power generation is improved.

Description

New energy power generation ultra-short term power prediction method and system based on iterative correction
Technical Field
The invention relates to the technical field of new energy power generation, in particular to a new energy power generation ultra-short term power prediction method and system based on iterative correction.
Background
The ultra-short-term power prediction of new energy power generation refers to power prediction in 15 minutes to 4 hours in the future, can provide decision support for new energy real-time scheduling, and can also provide reference for active power control of new energy stations and real-time trading of power markets. The ultra-short term power prediction method for new energy power generation based on the neural network is the most common ultra-short term power prediction method for new energy power generation at present, has strong fault tolerance, self-organization and self-adaption capability, and is very effective in solving the nonlinear problem. Because the variety of neural network methods is various, the LSTM neural network is taken as an example below to explain the LSTM-based new energy power generation ultra-short term power prediction method:
(1) LSTM neural network algorithm
LSTM is a time deep neural network, which comprises an input layer, a hidden layer, an output layer, a forgetting gate, an input gate, an output gate and an information flow representing long-term memory, and forms an inputxAnd a black box, i.e., cell, for the status output s, the specific structure is shown in fig. 1. Wherein the content of the first and second substances,x twhich is representative of the current input to the device,h t / h t-1to representt/t-the ratio of the hidden layer outputs at time 1,c t / c t-1to representt/t-a long-term memory state at time 1,f tin order to forget the output signal of the gate,i twhich represents the output signal of the input gate,c tfor the preliminary information to be currently entered into the long-term memory c,o tin order to output the output signal of the gate,h tis composed oftThe time instant implies the state value of the layer.
Figure 544248DEST_PATH_IMAGE001
In the formula: s (.) represents the activation function of the neuron;W xf representing the network weight value from the network input layer to the forgetting gate,W hf represents from the firstt-1 network coefficient weight b of output value of memory module to current forgetting gate at moment f A deviation vector representing the current forgetting gate.
Figure 573384DEST_PATH_IMAGE002
In the formula:W xi representing a network weight value from the network input layer to the current input gate;W hi represents from the firstt-1 the output value of the memory module at time to the network weight value of the current input gate;b i representing the offset vector of the current input gate.
Figure 345031DEST_PATH_IMAGE003
In the formula:W xo representing a network weight value from the network input layer to the current module;W ho represents from the firstt-1 memorizing the output value of the module to the network weight value of the current module at the moment;b o representing the deviation vector of the current output module.
Figure 92407DEST_PATH_IMAGE004
In the formula:W xc representing a network weight value from the network input layer to the current module;W hc represents from the firstt-1 memorizing the output value of the module to the network weight value of the current module at the moment;b c representing the deviation vector of the current memory module.
Figure 240491DEST_PATH_IMAGE005
In the formula:h tmeans for representing final long-short term memoryTo output of (c).
(2) LSTM neural network parameter design
The LSTM neural network-based ultra-short-term power prediction model for new energy power generation mainly needs to determine 5 parameters, namely input layer time step number, input layer dimension, number of hidden layers, dimension of each hidden layer and output variable dimension.
The input layer time step number is equal to the length of a variable time sequence used for carrying out ultra-short-term power prediction of new energy power generation. And inputting layer dimension, namely variable quantity, and when the variable quantity is multivariable, the input layer dimension is variable label quantity. The number of hidden layers, i.e. the number of LSTM layers, increases with the number of hidden layers, and the nonlinear fitting capability of the model increases with the number of training samples, but at the same time, the complexity of the model and the computation and time cost of training will also increase. The dimension of the hidden layer needs to be determined by comparing results through multiple tests, and is usually set to be three times of the number of variable labels. The output variable dimension is the number of output variables at each moment, and only one variable of the ultra-short term power of the new energy power generation needs to be output, so that the dimension is set to be 1.
(3) LSTM neural network input/output variables
Input variables are: actual power, wind speed, wind direction, temperature, sensible heat flux, latent heat flux, short wave radiation, long wave radiation, surface air pressure, large scale precipitation, convective precipitation, humidity, altitude, atmospheric density, and the like.
Output variables are: and generating power by using the new energy for 15min-4h in the future.
The ultra-short term power prediction by using the LSTM neural network has the following defects:
(1) the training speed is slow: more input variables and more complex calculation;
(2) the timeliness is not strong: the meteorological data based on the numerical model is difficult to update every 15 minutes;
(3) the device is easy to fall into a local extreme value;
(4) the degree of dependence on data is high: a large amount of historical samples are needed, and the newly-built new energy station cannot be applied.
In summary, the existing new energy power generation ultra-short-term power prediction method based on the neural network has the defects of low training speed, low real-time performance, easy falling into local minimum and the like, and needs a large number of historical samples, and particularly has a large error risk when the artificial neural network is used for carrying out ultra-short-term power prediction on a new energy station with incomplete data conditions or newly built new energy stations.
Disclosure of Invention
In order to solve the defects in the prior art, the invention provides a new energy power generation ultra-short-term power prediction method based on iterative correction, which comprises the following steps:
acquiring absolute errors and relative errors of short-term predicted power and theoretical generated power at a predicted time and a plurality of times before the current time, and absolute errors and relative errors between the short-term predicted power and actual generated power;
calculating a combined weighted absolute error and a combined weighted relative error based on the absolute error and the relative error between the short-term predicted power and theoretical generated power, the absolute error and the relative error between the short-term predicted power and actual generated power, and a weight coefficient obtained according to time sequence iteration;
and correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error or the combined weighted relative error to obtain the short-term predicted power value of the predicted time.
Preferably, the obtaining the short-term predicted power value at the predicted time based on the short-term predicted power value at the predicted time obtained by correcting the combined weighted absolute error or the combined weighted relative error includes:
when the relation between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time respectively meets a set condition, correcting the short-term predicted power value at the prediction time based on the combined weighted relative error to obtain an ultra-short-term predicted power value at the prediction time;
otherwise, correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error to obtain the ultra-short-term predicted power value of the predicted time.
Preferably, the relationship between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time satisfies a set condition as follows:
Figure 124134DEST_PATH_IMAGE006
and is and
Figure 66682DEST_PATH_IMAGE007
；
in the formula:δweighting the absolute error for the combination;c 1is a first parameter;
Figure 35775DEST_PATH_IMAGE008
to predict the time of dayjShort-term predicted power values of;θweighting the relative errors for the combination;c 2is the second parameter.
Preferably, the correcting the short-term predicted power value at the predicted time based on the combined weighted relative error obtains the short-term predicted power value at the predicted time, as shown in the following formula:
Figure 987550DEST_PATH_IMAGE009
in the formula:
Figure 725699DEST_PATH_IMAGE010
to predict the time of dayjThe ultra-short-term predicted power value;
Figure 104728DEST_PATH_IMAGE008
to predict the time of dayjShort-term predicted power values of;θthe relative errors are weighted for the combination.
Preferably, the correcting the short-term predicted power value at the predicted time based on the combined weighted absolute error to obtain the ultra-short-term predicted power value at the predicted time is as follows:
Figure 561117DEST_PATH_IMAGE011
in the formula:
Figure 51004DEST_PATH_IMAGE010
to predict the time of dayjThe ultra-short-term predicted power value;
Figure 909239DEST_PATH_IMAGE008
to predict the time of dayjShort-term predicted power values of;δthe absolute errors are weighted for the combination.
Preferably, the weight coefficients obtained by time-series iteration include:
acquiring a weight coefficient matrix of the current moment, and comparing the acquired ultra-short-term predicted power value of the current moment with an actual power generation power value:
when the ultra-short-term power predicted value is larger than the actual power generation power value, selecting historical time when the sum of the absolute error between the short-term predicted power and the theoretical power generation power and the absolute error between the short-term predicted power and the actual power generation power is larger than zero, and the sum of the relative error between the short-term predicted power and the theoretical power generation power and the relative error between the short-term predicted power and the actual power generation power is larger than zero in a weight coefficient matrix corresponding to the current time;
when the predicted value of the ultra-short-term power is smaller than the actual power generation power value, selecting historical time when the sum of the absolute error of the short-term predicted power and the theoretical power generation power and the absolute error between the short-term predicted power and the actual power generation power is smaller than zero and the sum of the relative error of the short-term predicted power and the theoretical power generation power and the relative error between the short-term predicted power and the actual power generation power is smaller than zero in a weight coefficient matrix corresponding to the current time;
halving the weight coefficient corresponding to the historical moment according to a bisection method to form a weight coefficient matrix B;
on the basis of the weight coefficient matrix B, multiplying each weight coefficient by the reciprocal of the sum of all weight coefficients to generate a weight coefficient matrix at the next moment;
wherein the weighting coefficients include: the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical generated power, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual generated power.
Preferably, after generating the weight coefficient matrix at the next time, the method further includes:
and sequentially judging whether the absolute error value between the theoretical power generation power value and the actual power generation power value corresponding to each historical moment is larger than or equal to the set proportion of the installed capacity of the new energy field station, if so, adjusting the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical power generation power at the historical moment, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual power generation power, and regenerating a weight coefficient matrix, otherwise, not adjusting.
Based on the same inventive concept, the invention also provides a new energy power generation ultra-short-term power prediction system based on iterative correction, which comprises:
the acquisition module is used for acquiring absolute errors and relative errors between short-term predicted power and theoretical generated power at a predicted time and a plurality of moments before the current time, and absolute errors and relative errors between the short-term predicted power and actual generated power;
the calculation module is used for calculating a combined weighted absolute error and a combined weighted relative error based on the absolute error and the relative error between the short-term predicted power and the theoretical generated power, the absolute error and the relative error between the short-term predicted power and the actual generated power and a weight coefficient obtained according to time sequence iteration;
and the prediction module is used for correcting the obtained short-term predicted power value of the prediction time based on the combined weighted absolute error or the combined weighted relative error to obtain the ultra-short-term predicted power value of the prediction time.
Preferably, the prediction module is specifically configured to:
when the relation between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time respectively meets a set condition, correcting the short-term predicted power value at the prediction time based on the combined weighted relative error to obtain an ultra-short-term predicted power value at the prediction time;
otherwise, correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error to obtain the ultra-short-term predicted power value of the predicted time.
Preferably, the system further comprises a generation module;
the generating module is used for iterating the obtained weight coefficients according to a time sequence;
correspondingly, the generating module is specifically configured to:
acquiring a weight coefficient matrix of the current moment, and comparing the acquired ultra-short-term predicted power value of the current moment with an actual power generation power value:
when the ultra-short-term power predicted value is larger than the actual power generation power value, selecting historical time when the sum of the absolute error between the short-term predicted power and the theoretical power generation power and the absolute error between the short-term predicted power and the actual power generation power is larger than zero, and the sum of the relative error between the short-term predicted power and the theoretical power generation power and the relative error between the short-term predicted power and the actual power generation power is larger than zero in a weight coefficient matrix corresponding to the current time;
when the predicted value of the ultra-short-term power is smaller than the actual power generation power value, selecting historical time when the sum of the absolute error of the short-term predicted power and the theoretical power generation power and the absolute error between the short-term predicted power and the actual power generation power is smaller than zero and the sum of the relative error of the short-term predicted power and the theoretical power generation power and the relative error between the short-term predicted power and the actual power generation power is smaller than zero in a weight coefficient matrix corresponding to the current time;
halving the weight coefficient corresponding to the historical moment according to a bisection method to form a weight coefficient matrix B;
on the basis of the weight coefficient matrix B, multiplying each weight coefficient by the reciprocal of the sum of all weight coefficients to generate a weight coefficient matrix at the next moment;
wherein the weighting coefficients include: the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical generated power, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual generated power.
Preferably, the system further includes a determining module, configured to sequentially determine, after generating a weight coefficient matrix at a next time, whether an absolute error value between a theoretical power generation power value and an actual power generation power value corresponding to each historical time is greater than or equal to a set proportion of the installed capacity of the new energy farm, if so, adjust a weight of an absolute error and a weight of a relative error between short-term predicted power and theoretical power generation power at the historical time, and a weight of an absolute error and a weight of a relative error between the short-term predicted power and actual power generation power, and regenerate the weight coefficient matrix, otherwise, not adjust.
Compared with the prior art, the invention has the beneficial effects that:
the technical scheme provided by the invention comprises the steps of obtaining the absolute error and the relative error between the short-term predicted power and the theoretical generated power at the predicted time and a plurality of times before the current time, and the absolute error and the relative error between the short-term predicted power and the actual generated power; calculating a combined weighted absolute error and a combined weighted relative error based on the absolute error and the relative error between the short-term predicted power and theoretical generated power, the absolute error and the relative error between the short-term predicted power and actual generated power, and a weight coefficient obtained according to time sequence iteration; and correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error or the combined weighted relative error to obtain the short-term predicted power value of the predicted time. The technical scheme utilizes the combined weighted error to iteratively correct the ultra-short-term predicted power of the new energy power generation at the predicted time, improves the ultra-short-term power prediction precision of the new energy power generation, has low requirement on the data volume of historical data, and is particularly suitable for new energy stations with incomplete data conditions or newly built new energy stations.
Drawings
FIG. 1 is a schematic structural diagram of a cell in an LSTM neural network;
fig. 2 is a flowchart of a method for predicting ultra-short term power of new energy power generation based on iterative correction according to an embodiment of the present invention;
fig. 3 is a detailed flowchart of a new energy power generation ultra-short term power prediction method based on iterative correction according to an embodiment of the present invention;
fig. 4 is a schematic diagram of an ultra-short term power prediction system for new energy power generation based on iterative correction according to an embodiment of the present invention.
Detailed Description
For a better understanding of the present invention, reference is made to the following description taken in conjunction with the accompanying drawings and examples.
The invention provides a new energy power generation ultra-short term power prediction method based on iterative correction, as shown in fig. 2, comprising the following steps:
s1, acquiring absolute errors and relative errors between short-term predicted power and theoretical generated power at a predicted time and a plurality of times before the current time, and absolute errors and relative errors between the short-term predicted power and actual generated power;
s2, calculating a combined weighted absolute error and a combined weighted relative error based on the absolute error and the relative error between the short-term predicted power and the theoretical generated power, the absolute error and the relative error between the short-term predicted power and the actual generated power, and a weight coefficient obtained by time sequence iteration;
s3 corrects the obtained short-term predicted power value at the predicted time based on the combined weighted absolute error or the combined weighted relative error to obtain an ultra-short-term predicted power value at the predicted time.
In this embodiment, the obtaining the short-term predicted power value at the predicted time based on the short-term predicted power value at the predicted time obtained by correcting the combined weighted absolute error or the combined weighted relative error includes:
when the relation between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time respectively meets a set condition, correcting the short-term predicted power value at the prediction time based on the combined weighted relative error to obtain an ultra-short-term predicted power value at the prediction time;
otherwise, correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error to obtain the ultra-short-term predicted power value of the predicted time.
In this embodiment, the weight coefficients obtained by time-series iteration include:
acquiring a weight coefficient matrix of the current moment, and comparing the acquired ultra-short-term predicted power value of the current moment with an actual power generation power value:
when the ultra-short-term power predicted value is larger than the actual power generation power value, selecting historical time when the sum of the absolute error between the short-term predicted power and the theoretical power generation power and the absolute error between the short-term predicted power and the actual power generation power is larger than zero, and the sum of the relative error between the short-term predicted power and the theoretical power generation power and the relative error between the short-term predicted power and the actual power generation power is larger than zero in a weight coefficient matrix corresponding to the current time;
when the predicted value of the ultra-short-term power is smaller than the actual power generation power value, selecting historical time when the sum of the absolute error of the short-term predicted power and the theoretical power generation power and the absolute error between the short-term predicted power and the actual power generation power is smaller than zero and the sum of the relative error of the short-term predicted power and the theoretical power generation power and the relative error between the short-term predicted power and the actual power generation power is smaller than zero in a weight coefficient matrix corresponding to the current time;
halving the weight coefficient corresponding to the historical moment according to a bisection method to form a weight coefficient matrix B;
on the basis of the weight coefficient matrix B, multiplying each weight coefficient by the reciprocal of the sum of all weight coefficients to generate a weight coefficient matrix at the next moment;
wherein the weighting coefficients include: the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical generated power, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual generated power.
As an alternative embodiment, after generating the weight coefficient matrix at the next time, the method further includes:
and sequentially judging whether the absolute error value between the theoretical power generation power value and the actual power generation power value corresponding to each historical moment is larger than or equal to the set proportion of the installed capacity of the new energy field station, if so, adjusting the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical power generation power at the historical moment, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual power generation power, and regenerating a weight coefficient matrix, otherwise, not adjusting.
Based on the content of the foregoing embodiment, as an alternative embodiment, as shown in fig. 3, a new energy generation ultra-short term power prediction method based on iterative correction is provided, including:
acquiring an absolute error between theoretical generating power and actual generating power of the new energy station based on actually-measured meteorological data and a theoretical conversion relation between meteorological-power of the new energy station at historical time;
respectively acquiring absolute errors and relative errors between the short-term predicted power of new energy power generation at historical time and theoretical power generation power and actual power generation power;
acquiring a combined weighted absolute error and a combined weighted relative error between the short-term predicted power of the new energy source power generation and the theoretical power generation power as well as between the short-term predicted power of the new energy source power generation and the actual power generation power at the historical moment according to the absolute error and the relative error between the short-term predicted power of the new energy source power generation and the theoretical power generation power as well as the actual power generation power at the historical moment and the installed capacity of the new energy source station;
and iteratively correcting the new energy power generation ultra-short-term predicted power at the predicted moment by utilizing the combined weighted absolute error and the combined weighted relative error.
As an optional embodiment, the obtaining an absolute error between theoretical generated power and actual generated power of the new energy station based on actually measured meteorological data and a theoretical conversion relationship between "meteorological-power" of the new energy station at the historical time includes:
based on the time before the current timeiActually measured wind speed (or irradiance) of wind power (or photovoltaic) station at historical moment) Value of
Figure 459169DEST_PATH_IMAGE012
(or
Figure 137275DEST_PATH_IMAGE013
) And wind speedv(or irradiance)R) -powerPConverting the relation:
P=f (v) (orP=f (R)）
Determining the time before the current time by the following formulaiTheoretical power generation power value of new energy station at historical momentP ai ：
P ai =f (v i ) (orP ai =f (R i )）
Determining theoretical generating power of the new energy station in historical time according to the following formulaP a Actual generated power of new energy station within data and historical timePAbsolute error value between dataε i ：
ε i =|P ai - P i |
In the above formula, the first and second carbon atoms are,iis the first before the current timeiAnd (4) historical time.
As an alternative embodiment, the obtaining absolute errors and relative errors between the short-term predicted power generated by the new energy and the theoretical generated power and between the short-term predicted power generated by the new energy and the actual generated power at the historical time respectively includes:
determining the time before the current time by the following formulaiNew energy power generation short-term power prediction value at historical moment
Figure 430853DEST_PATH_IMAGE014
Before the current timeiTheoretical power generation power value of new energy station at historical momentP ai Actual power generation power valueP i Absolute error value ofδ ai 、δ i ：
Figure 143594DEST_PATH_IMAGE015
Figure 864426DEST_PATH_IMAGE016
Wherein, the new energy power generation short-term power prediction value
Figure 295407DEST_PATH_IMAGE017
Ultra-short term power prediction
Figure 127097DEST_PATH_IMAGE018
Actual power generation power valuePAnd the measured wind speedv(or irradiance)R) The time resolution of the data is not less than 15min, and according to time scales of 00:15, 00:30, 00:45
Figure 694344DEST_PATH_IMAGE019
Actual generated powerPAnd the measured wind speedv(or irradiance)R）。
Determining short-term predicted power of new energy power generation in historical time according to the following formula
Figure 410296DEST_PATH_IMAGE020
Theoretical power generation power value of new energy station in data and historical timeP a Actual power generation power valuePRelative error value therebetweenθ ai 、θ i ：
Figure 62994DEST_PATH_IMAGE021
Figure 698375DEST_PATH_IMAGE022
In the above formula, the first and second carbon atoms are,iis the first before the current timeiAnd (4) historical time.
Through the calculation, the invention only needs to acquire the actually-measured meteorological data, the actual power data and the new energy short-term predicted power data of the new energy station at the historical moment, so that the invention has low data volume requirement on the historical data and good universality, and is particularly suitable for new energy stations with incomplete data conditions or newly-built new energy stations.
As an alternative embodiment, the acquiring, according to the absolute error, the relative error and the installed capacity of the new energy plant, the combined weighted absolute error and the combined weighted relative error between the short-term predicted power of new energy power generation at the historical time and the theoretical power generation power and the actual power generation power includes:
determining the combined absolute error between the new energy power generation short-term power prediction data at the historical moment and the actual power generation power data of the new energy station at the historical moment according to the following formulaδ：
Figure 120129DEST_PATH_IMAGE023
Determining the combined relative error between the short-term power prediction data of the new energy power generation at the historical moment and the actual power generation data of the new energy station at the historical moment according to the following formulaθ：
Figure 448342DEST_PATH_IMAGE024
In the above formula, the first and second carbon atoms are,iis the first before the current timeiN is the total number of historical moments before the current moment;δ ai 、δ i are respectively the firstiAbsolute error values of the short-term predicted power value of the new energy power generation at each historical moment, the theoretical power generation power value and the actual power generation power value;θ ai 、θ i are respectively the firstiShort-term predicted power of new energy power generation at historical momentRelative error values of the values and theoretical power generation power values and actual power generation power values;η ai 、η i respectively before the current timeiWeights of errors of short-term predicted power values of new energy power generation, theoretical power generation power values and actual power generation power values at historical moments, and a coefficient matrix A:
Figure 322757DEST_PATH_IMAGE025
satisfies the following conditions:
Figure 761828DEST_PATH_IMAGE026
and is
Figure 38089DEST_PATH_IMAGE027
In general terms, the amount of the solvent to be used,
Figure 537203DEST_PATH_IMAGE028
but before the current timeiTheoretical power generation power value of new energy station at historical momentP ai And the actual power generation power valueP i Absolute error value ofε i Greater than the installed capacity of the new energy field stationCOPA certain proportion ofλ：
ε i ≥λ.COP（0<λ<1;i=1,2,…,n）
Then will be
Figure 898915DEST_PATH_IMAGE029
Is assigned to
Figure 141677DEST_PATH_IMAGE030
Will be
Figure 538023DEST_PATH_IMAGE031
Is assigned to
Figure 208039DEST_PATH_IMAGE032
And will be
Figure 791467DEST_PATH_IMAGE033
And
Figure 837921DEST_PATH_IMAGE034
setting as 0:
Figure 415DEST_PATH_IMAGE037
the invention can share the error risk of single-point prediction, eliminate large deviation and improve the ultra-short-term power prediction precision of new energy power generation by combining real-time online iterative correction of weighted errors.
As an alternative embodiment, the iteratively correcting the new energy generation ultra-short term predicted power at the predicted time by using the combined weighted absolute error and the combined weighted relative error includes:
determining an initial weight coefficient matrix as follows:
Figure 584980DEST_PATH_IMAGE038
determining the predicted time byjUltra-short-term predicted power value of new energy power generation
Figure 690339DEST_PATH_IMAGE010
Firstly, extracting the short-term predicted power value of the new energy power generation at the moment
Figure 702157DEST_PATH_IMAGE008
Then considering the combined absolute errorδRelative error with combinationθThe size of (2). If it is
Figure 525757DEST_PATH_IMAGE006
And is and
Figure 648434DEST_PATH_IMAGE007
then
Figure 139458DEST_PATH_IMAGE009
Otherwise
Figure 56598DEST_PATH_IMAGE011
In the formula:δweighting the absolute error for the combination;c 1is a first parameter;
Figure 367494DEST_PATH_IMAGE008
to predict the time of dayjShort-term predicted power values of;θweighting the relative errors for the combination;c 2as the second parameter, the parameter is,c 1andc 2the criterion is set for empirical values, mainly for determining when to select absolute errors and when to select relative errors, and when the criterion is met, the weighted absolute errors are combinedδIs large, and should choose the combination weighted relative errorθ。
Predictive time-sequential rolling updates when a predicted time is obtainedjActual power generation power value of new energy stationP j Calculating the absolute error of the ultra-short-term power prediction of new energy power generation in real timeω j ：
Figure 28282DEST_PATH_IMAGE039
If it isω j >0, selecting
Figure 373813DEST_PATH_IMAGE040
Whereini=1,2,…,nCorresponding toiAt the moment, according to a dichotomy, a weighting coefficient is halved according to the following formula to form a coefficient matrix B;
Figure 461855DEST_PATH_IMAGE041
then, on the basis of the coefficient matrix B, each weight coefficient is multiplied by the reciprocal of the sum of all weight coefficients as follows to form a coefficient matrix C.
Figure 260047DEST_PATH_IMAGE042
Similarly, ifω j <0, selecting
Figure 724526DEST_PATH_IMAGE043
Whereini=1,2,…,nCorresponding toiAt the moment, the weighting coefficients are halved according to a dichotomy to form a coefficient matrix B, and then each weighting coefficient is multiplied by the reciprocal of the sum of all the weighting coefficients to form a coefficient matrix C.
And substituting the coefficient matrix C into the next step for rolling calculation, and correcting the new energy power generation ultra-short-term predicted power value at the predicted time.
The method utilizes the combined weighted error to iteratively correct the new energy power generation ultra-short-term prediction power at the prediction time, can provide decision support for the real-time scheduling of the new energy, and also can provide reference for the active power control of the new energy station and the real-time transaction of the electric power market.
In order to explain the effect of the method and the system for predicting the ultra-short term power of new energy power generation based on iterative correction, which are provided by the embodiment of the invention, a certain wind power plant with installed capacity of 60MW is taken as an example below, and the prediction effect of the invention is verified. Short-term predicted power data, actual generated power data, actual measured wind speed data and a theoretical conversion relation of wind speed-power are selected for verification, wherein the short-term predicted power data, the actual generated power data, the actual measured wind speed data and the theoretical conversion relation of wind speed-power are selected for seven days from 4 months to 4 months and 24 days in 2020, and the prediction effects are shown in table 1.
TABLE 1 prediction error index
Figure 924563DEST_PATH_IMAGE044
The results in table 1 show that various error indexes of the ultra-short-term predicted power based on the method are superior to those of the short-term predicted power and the ultra-short-term predicted power based on the neural network, and the method can effectively improve the ultra-short-term power prediction precision of new energy power generation.
In the method for predicting the ultra-short term power, the prior art focuses on weather-based classification (historical similarity days), and focuses on how to pick out a period similar to the current period in the historical data, the application utilizes the continuity of errors in the past hours (generally, 15min is one point, and a plurality of points are selected as the case may be) of the current moment, and focuses on how to reasonably distribute the influence of the continuity of the past errors on the future period, namely how to distribute weights, when to select relative errors or absolute errors, and how to roll and update iterations. According to the technical scheme, the ultra-short-term prediction power is determined by utilizing the error continuity, so that the method is suitable for a newly-built new energy station only by relying on historical data of a nearby time interval, and the method for determining the ultra-short-term prediction power by utilizing a similar day needs a large amount of historical data to be used as a support and is not suitable for the newly-built new energy station.
The essence of the ultra-short term forecast is to make corrections on the basis of the short term forecast, which is important in the application of the short term forecast of weather type deviations, e.g.TThe short-term prediction of the day results inTThe results of the 1-day prediction,Tthe day is a sunny day, howeverTWeather forecast handle of-1 dayTThe forecast of the day becomes rainy, which results in smaller short-term forecast power. Under the condition, the new energy power generation ultra-short-term power prediction method based on iterative correction can be utilized, and the method is based onTThe daily actual power effectively corrects the short-term power measurement deviation, and the accuracy of the ultra-short-term predicted power is improved. The method selects error continuation iteration to obtain the ultra-short-term power predicted value, and compared with the method of selecting history similar days to obtain the ultra-short-term predicted power in the prior art, the method has the advantages of less dependence on history data, simplicity, practicability and better universalityThe method is particularly suitable for new energy stations with incomplete data conditions or newly built new energy stations, decision support can be provided for real-time scheduling of new energy, reference can also be provided for active power control of the new energy stations and real-time trading of the electric power market, the method is simple and practical, the calculated amount is small, timeliness is high, implementation and engineering field deployment are easy, and the method has high operability and popularization and application values.
As shown in fig. 4, based on the same inventive concept, the present invention further provides an ultra-short term power prediction system for new energy power generation based on iterative correction, including:
the acquisition module is used for acquiring absolute errors and relative errors between short-term predicted power and theoretical generated power at a predicted time and a plurality of moments before the current time, and absolute errors and relative errors between the short-term predicted power and actual generated power;
the calculation module is used for calculating a combined weighted absolute error and a combined weighted relative error based on the absolute error and the relative error between the short-term predicted power and the theoretical generated power, the absolute error and the relative error between the short-term predicted power and the actual generated power and a weight coefficient obtained according to time sequence iteration;
and the prediction module is used for correcting the obtained short-term predicted power value of the prediction time based on the combined weighted absolute error or the combined weighted relative error to obtain the ultra-short-term predicted power value of the prediction time.
In an embodiment, the prediction module is specifically configured to:
when the relation between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time respectively meets a set condition, correcting the short-term predicted power value at the prediction time based on the combined weighted relative error to obtain an ultra-short-term predicted power value at the prediction time;
otherwise, correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error to obtain the ultra-short-term predicted power value of the predicted time.
In an embodiment, the correcting the short-term predicted power value at the predicted time based on the combined weighted relative error obtains the short-term predicted power value at the predicted time as shown in the following equation:
Figure 183506DEST_PATH_IMAGE009
in the formula:
Figure 468994DEST_PATH_IMAGE010
to predict the time of dayjThe ultra-short-term predicted power value;
Figure 737164DEST_PATH_IMAGE008
to predict the time of dayjShort-term predicted power values of;θthe relative errors are weighted for the combination.
In an embodiment, the correcting the short-term predicted power value at the predicted time based on the combined weighted absolute error obtains an ultra-short-term predicted power value at the predicted time, as shown in the following equation:
Figure 791708DEST_PATH_IMAGE011
in the formula:
Figure 221552DEST_PATH_IMAGE010
to predict the time of dayjThe ultra-short-term predicted power value;
Figure 994336DEST_PATH_IMAGE008
to predict the time of dayjShort-term predicted power values of;δthe absolute errors are weighted for the combination.
In an embodiment, the system further comprises a generation module;
the generating module is used for iterating the obtained weight coefficients according to a time sequence;
correspondingly, the generating module is specifically configured to:
acquiring a weight coefficient matrix of the current moment, and comparing the acquired ultra-short-term predicted power value of the current moment with an actual power generation power value:
when the ultra-short-term power predicted value is larger than the actual power generation power value, selecting historical time when the sum of the absolute error between the short-term predicted power and the theoretical power generation power and the absolute error between the short-term predicted power and the actual power generation power is larger than zero, and the sum of the relative error between the short-term predicted power and the theoretical power generation power and the relative error between the short-term predicted power and the actual power generation power is larger than zero in a weight coefficient matrix corresponding to the current time;
when the predicted value of the ultra-short-term power is smaller than the actual power generation power value, selecting historical time when the sum of the absolute error of the short-term predicted power and the theoretical power generation power and the absolute error between the short-term predicted power and the actual power generation power is smaller than zero and the sum of the relative error of the short-term predicted power and the theoretical power generation power and the relative error between the short-term predicted power and the actual power generation power is smaller than zero in a weight coefficient matrix corresponding to the current time;
halving the weight coefficient corresponding to the historical moment according to a bisection method to form a weight coefficient matrix B;
on the basis of the weight coefficient matrix B, multiplying each weight coefficient by the reciprocal of the sum of all weight coefficients to generate a weight coefficient matrix at the next moment;
wherein the weighting coefficients include: the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical generated power, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual generated power.
In an embodiment, the system further includes a determining module, configured to sequentially determine, after generating a weight coefficient matrix at a next time, whether an absolute error value between a theoretical power generation power value and an actual power generation power value corresponding to each historical time is greater than or equal to a set proportion of the installed capacity of the new energy farm, and if the absolute error value is greater than or equal to the set proportion, adjust a weight of an absolute error and a weight of a relative error between short-term predicted power and theoretical power generation power at the historical time, and adjust a weight coefficient matrix again according to a weight of an absolute error and a weight of a relative error between the short-term predicted power and actual power generation power, otherwise, not adjust the weight coefficient matrix.
The system firstly obtains absolute errors and relative errors of short-term predicted power and theoretical generated power at a predicted time and a plurality of times before the current time and absolute errors and relative errors between the short-term predicted power and actual generated power through an obtaining module;
secondly, calculating a combined weighted absolute error and a combined weighted relative error through a calculating module based on the absolute error and the relative error between the short-term predicted power and the theoretical generated power, the absolute error and the relative error between the short-term predicted power and the actual generated power, and a weight coefficient obtained according to time sequence iteration;
and finally, correcting the obtained short-term predicted power value of the predicted time based on the combined weighted absolute error or the combined weighted relative error through a prediction module to obtain the ultra-short-term predicted power value of the predicted time. The system utilizes the combined weighted error to iteratively correct the ultra-short term prediction power of the new energy power generation at the prediction time, improves the ultra-short term power prediction precision of the new energy power generation, has low requirement on the data volume of historical data, and is particularly suitable for new energy stations with incomplete data conditions or newly built new energy stations.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The present invention is not limited to the above embodiments, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention are included in the scope of the claims of the present invention which are filed as the application.

Claims (4)

1. A new energy power generation ultra-short-term power prediction method based on iterative correction is characterized by comprising the following steps:
acquiring absolute errors and relative errors of short-term predicted power and theoretical generated power at a predicted time and a plurality of times before the current time, and absolute errors and relative errors between the short-term predicted power and actual generated power;
calculating a combined weighted absolute error and a combined weighted relative error based on the absolute error and the relative error between the short-term predicted power and theoretical generated power, the absolute error and the relative error between the short-term predicted power and actual generated power, and a weight coefficient obtained according to time sequence iteration;
correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error or the combined weighted relative error to obtain an ultra-short-term predicted power value of the predicted time;
correspondingly, the obtaining the short-term predicted power value of the predicted time based on the short-term predicted power value of the predicted time obtained by correcting the combined weighted absolute error or the combined weighted relative error comprises:
when the relation between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time respectively meets a set condition, correcting the short-term predicted power value at the prediction time based on the combined weighted relative error to obtain an ultra-short-term predicted power value at the prediction time;
otherwise, correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error to obtain the ultra-short-term predicted power value of the predicted time;
wherein the relationship between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time respectively satisfies the set condition as follows:
Figure 299052DEST_PATH_IMAGE001
and is and
Figure 29110DEST_PATH_IMAGE002
；
in the formula:δweighting the absolute error for the combination;c 1is a first parameter;
Figure 979880DEST_PATH_IMAGE003
to predict the time of dayjShort-term predicted power values of;θweighting the relative errors for the combination;c 2is a second parameter;
correspondingly, the weight coefficients obtained by time-series iteration comprise:
acquiring a weight coefficient matrix of the current moment, and comparing the acquired ultra-short-term predicted power value of the current moment with an actual power generation power value:
when the ultra-short-term prediction power value is larger than the actual power generation power value, selecting historical time when the sum of the absolute error between the short-term prediction power and the theoretical power generation power and the absolute error between the short-term prediction power and the actual power generation power is larger than zero, and the sum of the relative error between the short-term prediction power and the theoretical power generation power and the relative error between the short-term prediction power and the actual power generation power is larger than zero in a weight coefficient matrix corresponding to the current time;
when the ultra-short-term prediction power value is smaller than the actual power generation power value, selecting historical time when the sum of the absolute error between the short-term prediction power and the theoretical power generation power and the absolute error between the short-term prediction power and the actual power generation power is smaller than zero, and the sum of the relative error between the short-term prediction power and the theoretical power generation power and the relative error between the short-term prediction power and the actual power generation power is smaller than zero in a weight coefficient matrix corresponding to the current time;
halving the weight coefficient corresponding to the historical moment according to a bisection method to form a weight coefficient matrix B;
on the basis of the weight coefficient matrix B, multiplying each weight coefficient by the reciprocal of the sum of all weight coefficients to generate a weight coefficient matrix at the next moment;
wherein the weighting coefficients include: the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical generated power, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual generated power;
after the generating of the weight coefficient matrix at the next time, the method further includes:
and sequentially judging whether the absolute error value between the theoretical power generation power value and the actual power generation power value corresponding to each historical moment is larger than or equal to the set proportion of the installed capacity of the new energy field station, if so, adjusting the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical power generation power at the historical moment, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual power generation power, and regenerating a weight coefficient matrix, otherwise, not adjusting.
2. The method of claim 1, wherein the correcting the short-term predicted power value for the predicted time based on the combined weighted relative error results in an ultra-short-term predicted power value for the predicted time as shown in the following equation:
Figure 504402DEST_PATH_IMAGE004
in the formula:
Figure 789890DEST_PATH_IMAGE005
to predict the time of dayjThe ultra-short-term predicted power value;
Figure 58060DEST_PATH_IMAGE003
to predict the time of dayjShort-term predicted power values of;θthe relative errors are weighted for the combination.
3. The method of claim 1, wherein the correcting the short-term predicted power value for the predicted time based on the combined weighted absolute error results in an ultra-short-term predicted power value for the predicted time as shown in the following equation:
Figure 190485DEST_PATH_IMAGE006
in the formula:
Figure 417067DEST_PATH_IMAGE005
to predict the time of dayjThe ultra-short-term predicted power value;
Figure 189851DEST_PATH_IMAGE003
to predict the time of dayjShort-term predicted power values of;δthe absolute errors are weighted for the combination.
4. A new energy power generation ultra-short-term power prediction system based on iterative correction is characterized by comprising:
the acquisition module is used for acquiring absolute errors and relative errors between short-term predicted power and theoretical generated power at a predicted time and a plurality of moments before the current time, and absolute errors and relative errors between the short-term predicted power and actual generated power;
the calculation module is used for calculating a combined weighted absolute error and a combined weighted relative error based on the absolute error and the relative error between the short-term predicted power and the theoretical generated power, the absolute error and the relative error between the short-term predicted power and the actual generated power and a weight coefficient obtained according to time sequence iteration;
the prediction module is used for correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error or the combined weighted relative error to obtain the ultra-short-term predicted power value of the predicted time;
the prediction module is specifically configured to:
when the relation between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time respectively meets a set condition, correcting the short-term predicted power value at the prediction time based on the combined weighted relative error to obtain an ultra-short-term predicted power value at the prediction time;
otherwise, correcting the short-term predicted power value of the predicted time based on the combined weighted absolute error to obtain the ultra-short-term predicted power value of the predicted time;
wherein the relationship between the combined weighted absolute error and the combined weighted relative error and the short-term predicted power value at the prediction time respectively satisfies the set condition as follows:
Figure 746865DEST_PATH_IMAGE001
and is and
Figure 718232DEST_PATH_IMAGE002
；
in the formula:δweighting the absolute error for the combination;c 1is a first parameter;
Figure 584557DEST_PATH_IMAGE003
to predict the time of dayjShort-term predicted power values of;θweighting the relative errors for the combination;c 2is a second parameter;
the system further comprises a generation module;
the generating module is used for iterating the obtained weight coefficients according to a time sequence;
correspondingly, the generating module is specifically configured to:
acquiring a weight coefficient matrix of the current moment, and comparing the acquired ultra-short-term predicted power value of the current moment with an actual power generation power value:
when the ultra-short-term prediction power value is larger than the actual power generation power value, selecting historical time when the sum of the absolute error between the short-term prediction power and the theoretical power generation power and the absolute error between the short-term prediction power and the actual power generation power is larger than zero, and the sum of the relative error between the short-term prediction power and the theoretical power generation power and the relative error between the short-term prediction power and the actual power generation power is larger than zero in a weight coefficient matrix corresponding to the current time;
when the ultra-short-term prediction power value is smaller than the actual power generation power value, selecting historical time when the sum of the absolute error between the short-term prediction power and the theoretical power generation power and the absolute error between the short-term prediction power and the actual power generation power is smaller than zero, and the sum of the relative error between the short-term prediction power and the theoretical power generation power and the relative error between the short-term prediction power and the actual power generation power is smaller than zero in a weight coefficient matrix corresponding to the current time;
halving the weight coefficient corresponding to the historical moment according to a bisection method to form a weight coefficient matrix B;
on the basis of the weight coefficient matrix B, multiplying each weight coefficient by the reciprocal of the sum of all weight coefficients to generate a weight coefficient matrix at the next moment;
wherein the weighting coefficients include: the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical generated power, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual generated power;
the system also comprises a judging module, which is used for sequentially judging whether the absolute error value between the theoretical power generation power value and the actual power generation power value corresponding to each historical moment is larger than or equal to the set proportion of the installed capacity of the new energy field station after generating the weight coefficient matrix of the next moment, if so, adjusting the weight of the absolute error and the weight of the relative error between the short-term predicted power and the theoretical power generation power at the historical moment, and the weight of the absolute error and the weight of the relative error between the short-term predicted power and the actual power generation power, and regenerating the weight coefficient matrix, otherwise, not adjusting.
CN202011227869.1A 2020-11-06 2020-11-06 New energy power generation ultra-short term power prediction method and system based on iterative correction Active CN112036672B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011227869.1A CN112036672B (en) 2020-11-06 2020-11-06 New energy power generation ultra-short term power prediction method and system based on iterative correction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011227869.1A CN112036672B (en) 2020-11-06 2020-11-06 New energy power generation ultra-short term power prediction method and system based on iterative correction

Publications (2)

Publication Number Publication Date
CN112036672A CN112036672A (en) 2020-12-04
CN112036672B true CN112036672B (en) 2021-04-20

Family

ID=73572750

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011227869.1A Active CN112036672B (en) 2020-11-06 2020-11-06 New energy power generation ultra-short term power prediction method and system based on iterative correction

Country Status (1)

Country Link
CN (1) CN112036672B (en)

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101998553B1 (en) * 2012-08-01 2019-07-10 한국전력공사 Prediction Method of Short-Term Wind Speed and Wind Power and Power Supply Line Voltage Prediction Method Therefore
CN102914969B (en) * 2012-09-28 2014-12-10 南方电网科学研究院有限责任公司 Comprehensive error correction method of short-period wind power prediction system
JP6641849B2 (en) * 2015-10-01 2020-02-05 日本製鉄株式会社 Power prediction method
CN111832786A (en) * 2019-04-23 2020-10-27 中国电力科学研究院有限公司 Photovoltaic power generation ultra-short term power prediction method and device based on sliding error
CN110263971A (en) * 2019-05-14 2019-09-20 西安理工大学 Super short-period wind power combination forecasting method based on support vector machines

Also Published As

Publication number Publication date
CN112036672A (en) 2020-12-04

Similar Documents

Publication Publication Date Title
CN111008728B (en) Prediction method for short-term output of distributed photovoltaic power generation system
Han et al. Mid-to-long term wind and photovoltaic power generation prediction based on copula function and long short term memory network
Behera et al. Solar photovoltaic power forecasting using optimized modified extreme learning machine technique
Ramsami et al. A hybrid method for forecasting the energy output of photovoltaic systems
Tao et al. Forecasting power output for grid-connected photovoltaic power system without using solar radiation measurement
CN103117546B (en) A kind of Ultrashort-term slide prediction method for wind power
CN108053048A (en) A kind of gradual photovoltaic plant ultra-short term power forecasting method of single step and system
CN111144644B (en) Short-term wind speed prediction method based on variation variance Gaussian process regression
CN109088407B (en) Power distribution network state estimation method based on deep belief network pseudo-measurement modeling
CN110601250B (en) Method for evaluating stability of static voltage of access point based on new energy power station
CN113537582B (en) Photovoltaic power ultra-short-term prediction method based on short-wave radiation correction
WO2023216576A1 (en) Photovoltaic power generation short-term power prediction method and system
CN114330934A (en) Model parameter self-adaptive GRU new energy short-term power generation power prediction method
Karabiber et al. Short term PV power estimation by means of extreme learning machine and support vector machine
Yang et al. Kalman filter photovoltaic power prediction model based on forecasting experience
Quang et al. Applying Artificial Intelligence in Forecasting the Output of Industrial Solar Power Plant in Vietnam
Alharbi et al. Short-term wind speed and temperature forecasting model based on gated recurrent unit neural networks
CN113991711A (en) Capacity configuration method for energy storage system of photovoltaic power station
CN111967660B (en) Ultra-short-term photovoltaic prediction residual error correction method based on SVR
CN111461297B (en) Solar irradiation quantity optimization prediction algorithm based on MPC and ELM neural network
CN113111592A (en) Short-term wind power prediction method based on EMD-LSTM
CN111697560B (en) Method and system for predicting load of power system based on LSTM
CN112036672B (en) New energy power generation ultra-short term power prediction method and system based on iterative correction
CN116167508A (en) Short-term photovoltaic output rapid prediction method and system based on meteorological factor decomposition
CN115907131A (en) Method and system for building electric heating load prediction model in northern area

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
GR01 Patent grant
GR01 Patent grant