CN113822492A - Short-term power load prediction method and device and readable storage medium - Google Patents

Short-term power load prediction method and device and readable storage medium Download PDF

Info

Publication number
CN113822492A
CN113822492A CN202111183431.2A CN202111183431A CN113822492A CN 113822492 A CN113822492 A CN 113822492A CN 202111183431 A CN202111183431 A CN 202111183431A CN 113822492 A CN113822492 A CN 113822492A
Authority
CN
China
Prior art keywords
power load
load data
prediction
short
sequence
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
Application number
CN202111183431.2A
Other languages
Chinese (zh)
Inventor
孟学艺
胡蕴韬
王为国
付雪影
吕洪光
王志浩
张嘉伟
王同同
郭长虹
张义国
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.)
Bincheng Power Supply Co Of State Grid Shandong Electric Power Co
Binzhou Power Supply Co of State Grid Shandong Electric Power Co Ltd
Original Assignee
Bincheng Power Supply Co Of State Grid Shandong Electric Power Co
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 Bincheng Power Supply Co Of State Grid Shandong Electric Power Co filed Critical Bincheng Power Supply Co Of State Grid Shandong Electric Power Co
Priority to CN202111183431.2A priority Critical patent/CN113822492A/en
Publication of CN113822492A publication Critical patent/CN113822492A/en
Pending legal-status Critical Current

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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/067—Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • Theoretical Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Development Economics (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Health & Medical Sciences (AREA)
  • Educational Administration (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention provides a short-term power load prediction method, equipment and a readable storage medium, which configure a power load data sample sequence; configuring a power load data sample sequence to a membership function; combining an improved fuzzy averaging function calculation mode of an inverse reasoning theory with an optimal subset regression algorithm to deduce a power load short-term prediction model; firstly, defining a continuation sequence of a fuzzy averaging function as a free variable based on an optimal subset regression algorithm; independent variables are freely combined; and respectively establishing a linear regression equation by the combined sequences and the dependent variable, and screening a short-term load prediction model of the fuzzy average generation function from all the regression equations according to a preset screening standard to obtain a corresponding load prediction value. The hidden useful information in the actual load data is fully mined, and the problem of failure of the adjacent data of the periodic algorithm in the traditional model is effectively solved.

Description

Short-term power load prediction method and device and readable storage medium
Technical Field
The present invention relates to the field of power load prediction technologies, and in particular, to a short-term power load prediction method, device, and readable storage medium.
Background
The short-term power load prediction refers to prediction of a period of time in the future in the operation process of the power system, the short-term power load prediction is an important daily work of a power system scheduling operation department, a power utilization service department and a power grid planning department, and prediction data can provide a basis for making a power generation plan, a power transmission scheme and power grid construction.
With continuous deepening of power market reform, advanced and practical short-term power load prediction can provide strong support for power production and economic optimization scheduling of a power distribution network, meanwhile, the method can provide basis for ordering and selling power decisions of a market trading main system, and has strong practical application value under new situation.
At present, power load prediction is an important component of a power system, and for each station area in a power grid, daily load prediction with higher precision is provided for the scheduling of a control end at the upper stage, so that the scheduling of the control end at the upper stage can reasonably arrange the rotating standby capacity and the cold standby capacity when a power generation plan is made, the number of start and stop times of a unit is reduced, the daily transaction electric quantity is reduced, the power cost and the power price are reduced while the power consumption needs of users are met, and for the power grid of the power grid, daily load prediction data is an important basis for checking the safety of the power grid. The load prediction data can also meet the power grid maintenance plan.
At present, extensive research is carried out on the problem of short-term load prediction, and the current prediction method is generally a modern prediction method based on machine learning as a theoretical basis and a classical prediction method based on a time series prediction principle. The prediction method has poor actual operability and high requirement on the completeness of the data sample, is easily restricted by data acquisition during prediction, and can cause distortion of a short-term power load prediction result and seriously influence subsequent scheduling work if the data sample has deviation or does not meet the prediction method.
Disclosure of Invention
In order to improve the accuracy of short-term power load prediction and aim at the characteristics of periodicity, random fluctuation and the like of a power load sequence, the invention provides a short-term power load prediction method based on the combination of a fuzzy averaging function and optimal subset regression, which comprises the following steps:
configuring a power load data sample sequence;
configuring a power load data sample sequence to a membership function;
combining an improved fuzzy averaging function calculation mode of an inverse reasoning theory with an optimal subset regression algorithm to deduce a power load short-term prediction model;
firstly, defining a continuation sequence of a fuzzy averaging function as a free variable based on an optimal subset regression algorithm;
independent variables are freely combined;
and respectively establishing a linear regression equation by the combined sequences and the dependent variable, and screening a short-term load prediction model of the fuzzy average generation function from all the regression equations according to a preset screening standard to obtain a corresponding load prediction value.
It should be further noted that the step of configuring the power load data to the membership function in the membership function is as follows:
Figure BDA0003298198140000021
wherein:μaccording to the presetting of the past actual value;
if the time-ordered power load data sample sequence is a periodic sequence, the membership function is:
Figure BDA0003298198140000031
wherein: l represents the period length of the sequence; r is a constant determined empirically or by trial calculation;
if the action of the data in the power load data sample sequence on the prediction point gradually decreases with the distance and shows the periodicity of the power load data, the membership function is as follows:
Figure BDA0003298198140000032
it should be further noted that, for the time-ordered power load data sample sequence x (t), the fuzzy averaging function is defined according to the following formula:
Figure BDA0003298198140000033
wherein: 1, 2, …, l; l is more than or equal to 1 and less than or equal to m; n islINT (n/l); m ═ INT (n/2) or INT (n/3); l is the period corresponding to the fuzzy averaging function; INT is data rounding.
It is further noted that the fuzzy averaging function is:
Figure BDA0003298198140000034
wherein: rl＝n-nl·l；RlThe number of the residual items with the number of the samples being n;
will R toiThe + i term is used as the starting point of calculating the fuzzy average generating function and is calculated to the last term according to a preset interval, wherein n is-RlIs a multiple of the period l.
It should be further noted that, the fuzzy average generation function on a power load data sample sequence is expanded to the whole prediction interval, and periodic continuation prediction is performed;
the periodic continuation prediction mode is as follows:
Figure BDA0003298198140000041
wherein: 1, 2, …, n; mod represents a numerical remainder; obtaining a periodic continuation matrix of the fuzzy average generation function:
G＝(gi，j)n×l，gi，j≡gl(t) (7)
Figure BDA0003298198140000042
wherein: nxl is the order number of the period continuation matrix;
Figure BDA0003298198140000043
is the fuzzy averaging function sequence generated when l is 1;
Figure BDA0003298198140000044
show taking in sequence
Figure BDA0003298198140000045
One of them;
Figure BDA0003298198140000046
show taking in sequence
Figure BDA0003298198140000047
One, the rest, and so on.
It should be further noted that the method for screening out the short-term load prediction model of the fuzzy averaging function to obtain the corresponding load prediction value includes:
1) performing difference processing on an original sample sequence x (t), and obtaining two difference sequences by means of total calculation;
x(1)(t)＝{Δx(1)，Δx(2)，…，Δx(n-1)} (9)
x(2)(t)＝{Δ2x(1)，Δ2x(2)，…，Δ2x(n-1)} (10)
2) setting mu to be 0.01 based on the membership function;
calculating fuzzy average generation function sequences of three adjacent period power load data sample sequences according to the formula (5), and respectively recording the fuzzy average generation function sequences as
Figure BDA0003298198140000051
And
Figure BDA0003298198140000052
then, the corresponding cycle continuation sequences are respectively calculated and obtained by the formula (6)
Figure BDA0003298198140000053
And
Figure BDA0003298198140000054
3) calculating an accumulated sequence of first order difference sequences:
Figure BDA0003298198140000055
wherein: gl (3)(1)＝x(1)；t＝2，3，…，n；l＝1，2，…，m；
Generating 4m fuzzy average generating function continuation sequences gl (0)(t)、gl (1)(t)、gl (2)(t) and gl (3)(t)；
4) Quote the extension sequence g that the double scoring rule pair calculatesl (0)(t)、gl (1)(t)、gl (2)(t) and gl (3)(t) screening;
the rationale for the double scoring criterion is as follows:
CSC＝M1+M2 (12)
wherein: m1Scoring the quantity; m2 Trend towardsGrading;
the quantitative score is defined as:
Figure BDA0003298198140000056
wherein: r2Is a complex correlation coefficient; n is the length of the sample sequence; qKIs the sum of the squares of the residuals of the prediction model; qXIs the sum of the squares of the total deviations of the prediction model;
and the trend score is according to the minimum discrimination information statistic criterion:
M2＝2[T1+(n-1)·ln(n-1)-T2-T3] (14)
wherein:
Figure BDA0003298198140000061
g is the number of categories of the predicted trend;
when the CSC value is maximum, the corresponding prediction model has the best effect, so the maximum CSC value is used as the standard for screening the optimal subset;
5) respectively carrying out unitary regression calculation on the obtained extension sequence of the fuzzy average generation function and the original power load data sample sequence, and solving the CSC value of the extension sequence;
then chi fang test is carried out to lead the CSC value to be larger than chi2αThe corresponding sequence of (2) is defined as a prediction factor of a prediction equation, and the number of the prediction factor is recorded as k, namely k free variables are obtained;
according to the Newton's binomial theorem, 2 can be calculatedk-1 permutation combination, each combination to constitute a subset of the load prediction equation;
6) selecting the prediction factors obtained by rough selection;
2 obtained by calculationk-1 subset respectively constructing a multiple linear regression equation of the original power load data sample sequence;
solving the CSC value of the regression result according to the double-scoring criterion;
from 2k1, selecting the subset with the maximum CSC value as the optimal subset of the prediction equation;
7) assuming that the optimal subset contains p independent variables, the obtained short-term load prediction model based on the fuzzy averaging function is as follows:
Figure BDA0003298198140000071
if load prediction of w steps is to be completed, the sequence g is processedi(t) (i ═ 1, 2, …, p) we refer to equation (6) to make w steps cycle prolongation, then it is substituted into the derived short-term load prediction model to get the corresponding load prediction value.
It should be further noted that, the accuracy of each prediction model is quantitatively evaluated, the error evaluation function adopts a Relative Error (RE), a Mean Absolute Percentage Error (MAPE) and a Root Mean Square Error (RMSE), and the expressions are respectively:
Figure BDA0003298198140000072
Figure BDA0003298198140000073
Figure BDA0003298198140000074
wherein: n is the number of test samples; x is the number oftThe actual load value at the t-th moment;
Figure BDA0003298198140000075
is the predicted load value at the t-th time.
It should be further noted that the configuring the power load data sample sequence mode includes:
presetting at least three power load data acquisition cycles;
acquiring a plurality of power load data in each power load data acquisition period;
calculating the average value and the standard deviation of the power load data in the power load data acquisition period;
setting a power load data comparison parameter range, wherein the power load data comparison parameter range is the average value of power load data plus or minus 2 multiplied by the standard deviation of the power load data;
and comparing each power load data in a preset time period with the power load data comparison parameter range, if the power load data comparison parameter range is exceeded, deleting the current power load data, and calling the power load data from the subsequent period of the power load data acquisition period for supplement.
The invention also provides equipment for realizing the short-term power load forecasting method, which comprises the following steps:
a memory for storing a computer program and a short-term power load prediction method;
a processor for executing the computer program and the short term power load prediction method to realize the steps of the short term power load prediction method.
The present invention also provides a readable storage medium having a short-term power load prediction method, the readable storage medium having stored thereon a computer program, the computer program being executed by a processor to implement the steps of the short-term power load prediction method.
According to the technical scheme, the invention has the following advantages:
in the short-term power load prediction method provided by the invention, when the prediction model combining the fuzzy averaging function and the optimal subset regression is applied to short-term power load prediction, the interference of external factors can be reduced, for example, the influence factors of loads such as weather, date types, economic levels and the like do not need to be considered, and the method has the advantages of simplicity and convenience in implementation, strong operability and good prediction effect, so that the model has higher practical application value.
The effective load prediction curve of the prediction model built by the invention is closer to the actual load, and the fitting to the fluctuation direction of the load data is more accurate. Meanwhile, 3 typical prediction error quantization indexes show that the FMGF-OSR model provided by the invention is obviously better than the MGF-OSR model. The FMGF-OSR model integrates the advantages of fuzzy averaging function, inverse reasoning theory and optimal subset regression, fully excavates useful information hidden in actual load data, effectively solves the problem of failure of adjacent data of a periodic algorithm in a traditional model, and considers the great influence of recent load data on a prediction result and the influence of periodic change of the load data in different time scales.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings used in the description will be briefly introduced, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained based on these drawings without creative efforts.
FIG. 1 is a flow chart of a short term power load prediction method;
FIG. 2 is a comparison of load prediction results for three prediction models.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to fig. 1 in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The elements and algorithm steps of each example described in the embodiments disclosed in the short term power load prediction method provided by the present invention can be implemented in electronic hardware, computer software, or a combination of both, and in the above description the components and steps of each example have been generally described in terms of functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the method for short-term power load prediction provided by the present invention, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.
The short-term power load prediction method provided by the invention specifically comprises the following steps:
s1, configuring a power load data sample sequence;
presetting at least three power load data acquisition cycles;
acquiring a plurality of power load data in each power load data acquisition period;
calculating the average value and the standard deviation of the power load data in the power load data acquisition period;
setting a power load data comparison parameter range, wherein the power load data comparison parameter range is the average value of power load data plus or minus 2 multiplied by the standard deviation of the power load data;
and comparing each power load data in a preset time period with the power load data comparison parameter range, if the power load data comparison parameter range is exceeded, deleting the current power load data, and calling the power load data from the subsequent period of the power load data acquisition period for supplement.
By processing the power load data of each cycle, the prediction accuracy can be effectively improved.
S2, configuring the power load data sample sequence to a membership function;
s3, combining an inverse reasoning theory improved fuzzy averaging function calculation mode with an optimal subset regression algorithm to deduce a power load short-term prediction model;
s4, defining the continuation sequence of the fuzzy average generation function as a free variable based on the optimal subset regression algorithm;
s5, independent variables are freely combined;
and S6, establishing linear regression equations respectively for the combined sequences and the dependent variables, and screening short-term load prediction models of the fuzzy averaging functions from all the regression equations according to preset screening standards to obtain corresponding load prediction values.
The invention provides a short-term power load prediction model based on the combination of a fuzzy average generation function and optimal subset regression aiming at the characteristics of periodicity, random fluctuation and the like of a power load sequence. The method comprises the steps of firstly introducing a fuzzy averaging function algorithm into the load prediction field, simultaneously improving the construction process of the fuzzy averaging function by applying an inverse push theory, then combining the fuzzy averaging function algorithm with an optimal subset regression algorithm, establishing a short-term load prediction model, and finally predicting by using the model.
The invention combines the construction principle of the homogenesis function with the fuzzy theory, deduces the membership degree which can reflect different characteristics of the sequence sample, and provides the definition process of the fuzzy homogenesis function.
In order to solve the actual problem, the numerical value close to the predicted point can play a larger role while making full use of the past information. Thus, exponentially decreasing membership functions are designed, i.e.
Figure BDA0003298198140000111
Wherein: μ is set in advance according to the degree of importance given to the past actual value.
If the time-ordered power load data sample sequence is a periodic sequence, the membership function is:
Figure BDA0003298198140000112
wherein: l represents the period length of the sequence; r is a constant determined empirically or by trial calculation.
If the action of the data in the power load data sample sequence on the prediction point gradually decreases with the distance and shows the periodicity of the power load data, the membership function is as follows:
Figure BDA0003298198140000113
for the time series sample x (t), the fuzzy averaging function is defined as follows:
Figure BDA0003298198140000121
wherein: 1, 2, …, l; l is more than or equal to 1 and less than or equal to m; n islINT (n/l); m ═ INT (n/2) or INT (n/3); l is the period corresponding to the fuzzy averaging function; INT is data rounding.
The fuzzy averaging function is a periodic function generated by assigning each numerical value in the sequence with a certain membership degree and then solving an average value according to different intervals. Therefore, in the process of constructing the fuzzy average power function, due to the problem of the periodic algorithm, it cannot be guaranteed that tail data of the power load data sample sequence are considered in the fuzzy average power function continuation sequence corresponding to each period. The data at the end of the power load data sample sequence has a significant impact on the actual prediction. Aiming at the problems, the construction process of the fuzzy averaging function is improved by utilizing an inverse reasoning theory so as to ensure that the function of data at the tail part of the sample sequence in actual prediction is realized. The fuzzy average generating function after the improvement of the inverse reasoning theory has the formula:
Figure BDA0003298198140000122
wherein: rl＝n-nl·l；RlThe number of the residual items with the number of the samples being n; the other parameters are as defined above.
It can be known that the computing order of the sample sequence is optimized by using the inverse reasoning theory in the process of constructing the fuzzy averaging function, namely, the RthlThe + i term is used as the starting point for calculating the fuzzy averaging function, and then is calculated at different intervals until the last term, where n-RlThe period is multiple of 1, and the processing ensures that the utility of the tail data of the power load data sample sequence can be realized.
Extending the fuzzy average function in a period to the whole interval, also called periodic continuation:
Figure BDA0003298198140000131
wherein: 1, 2, …, n; mod represents a numerical remainder. Then, a periodic extension matrix of the fuzzy average generation function is obtained through calculation:
G＝(gi，j)n×l，gi，j≡gl(t) (7)
Figure BDA0003298198140000132
wherein: n × l is the order of the period extension matrix.
Figure BDA0003298198140000133
Is the fuzzy averaging function sequence generated when l is 1;
Figure BDA0003298198140000134
show taking in sequence
Figure BDA0003298198140000135
One of them;
Figure BDA0003298198140000136
show taking in sequence
Figure BDA0003298198140000137
One, the rest, and so on.
Aiming at the randomness and periodicity of short-term power load data and the complexity of influence factors, the method is combined with Optimal Subset Regression (OSR) to deduce a power load short-term prediction model on the basis of improving a fuzzy averaging function calculation method by using an inverse push theory. The basic idea of the optimal subset regression algorithm is that a fuzzy average generation function continuation sequence is defined as a free variable, then the free combination is carried out on the independent variable, then a linear regression equation is established between the combined sequence and a dependent variable, and finally an equation with the best effect is screened out from all regression equations according to a certain screening standard. The specific modeling process is as follows:
1) in order to realize the function of fitting the high-frequency components of the original sequence, the original sample sequence x (t) is subjected to differential processing, and two differential sequences are obtained in total.
x(1)(t)＝{Δx(1)，Δx(2)，…，Δx(n-1)} (9)
x(2)(t)＝{Δ2x(1)，Δ2x(2)，…，Δ2x(n-1)} (10)
2) Based on the membership function, when μ is not 0.01, the fuzzy averaging function sequences of the above three sequences are calculated according to the formula (5) and are respectively recorded as
Figure BDA0003298198140000141
And
Figure BDA0003298198140000142
then, the corresponding cycle continuation sequences g can be respectively calculated by the formula (6)l (0)(t)、gl (1)(t) and gl (2))(t)。
3) To fit the trend of the power load data sample sequence, the accumulated sequence of the first order difference sequence is further calculated:
Figure BDA0003298198140000143
wherein: gl (3)(1) X (1); t is 2, 3, …, n; l is 1, 2, …, m. It can be known that, because l can take m different values, the number of elements of different fuzzy averaging function sequences is different, and finally about 4m fuzzy averaging function extension sequences g are generatedl (0)(t)、gl (1)(t)、gl (2)(t) and gl (3)(t)。
4) And (5) screening the continuation sequences obtained by the calculation by referring to a double-scoring criterion.
The rationale for the double scoring criterion is as follows:
CSC＝M1+M2 (12)
wherein: m1Scoring the quantity; m2Is a trend score.
The quantitative score is defined as:
Figure BDA0003298198140000151
wherein: r2Is a complex correlation coefficient; n is the length of the sample sequence; qKIs the sum of the squares of the residuals of the prediction model; qXIs the sum of the squares of the total deviations of the predictive model.
And the trend score is according to the minimum discrimination information statistic criterion:
M2＝2[T1+(n-1)·ln(n-1)-T2-T3] (14)
wherein:
Figure BDA0003298198140000152
g is the number of categories of predicted trends.
The double-scoring criterion aims to enable the prediction model to have higher fitting precision and to judge the trend of the data more accurately. It can be seen that the prediction model corresponding to the maximum CSC value has the best effect, so the maximum CSC value is used as the standard for screening the optimal subset.
5) Respectively carrying out unitary regression calculation on the obtained extension sequence of the fuzzy average generation function and the power load data sample sequence, and solving the CSC value of the extension sequence; then chi fang test is carried out to lead the CSC value to be larger than chi2 αThe corresponding sequence of (a) is defined as the predictor of the prediction equation, and the number of the predictor is recorded as k, namely k free variables are obtained. According to the Newton's binomial theorem, 2 can be calculatedk-1 permutation combination, each combination constituting a subset of the load prediction equation.
6) And (4) refining the prediction factors obtained by rough selection. 2 obtained by calculationk-1 subset respectively constructing a multiple linear regression equation of the power load data sample sequence; then, solving the CSC value of the regression result again according to a double-scoring criterion; finally, from 2kAnd 1, selecting the subset with the maximum CSC value as the optimal subset of the prediction equation.
7) Assuming that the optimal subset contains p independent variables, the obtained short-term load prediction model based on the fuzzy averaging function is as follows:
Figure BDA0003298198140000161
if load prediction of w steps is to be completed, the sequence g is carried outi(t) (i ═ 1, 2, …, p) we refer to equation (6) to extend the period of w steps, then we can get the corresponding predicted value of load by substituting it into the derived short-term load prediction model.
In the invention, in order to effectively and comprehensively carry out quantitative evaluation on the accuracy of each prediction model, the error evaluation function selects a Relative Error (RE), a Mean Absolute Percent Error (MAPE) and a Root Mean Square Error (RMSE), and the expressions are respectively as follows:
Figure BDA0003298198140000171
Figure BDA0003298198140000172
Figure BDA0003298198140000173
wherein: n is the number of test samples; x is the number oftThe actual load value at the t-th moment;
Figure BDA0003298198140000174
is the predicted load value at the t-th time.
The load prediction values obtained by using the MATLAB programming and respectively adopting 2 prediction models are shown in FIG. 2.
As can be seen from FIG. 2, the load prediction curve obtained according to the FMGF-OSR model is closer to the actual load data, and the prediction precision is highest. Next, the superiority of the FMGF-OSR model is further described in detail in conjunction with the error quantization index of the prediction result, and specific statistical results are shown in tables 1 and 2.
TABLE 1 comparison of predicted and true load values
Figure BDA0003298198140000175
Figure BDA0003298198140000181
TABLE 2 prediction error comparison
％
Figure BDA0003298198140000182
Note: max and Min are respectively the maximum relative error and the minimum relative error of the prediction model.
As can be seen from tables 1 and 2, the effective load prediction curve of the established prediction model is closer to the actual load, and the fitting to the fluctuation direction of the load data is more accurate. Meanwhile, 3 typical prediction error quantization indexes show that the FMGF-OSR model provided by the invention is obviously better than the MGF-OSR model. The FMGF-OSR model integrates the advantages of fuzzy averaging function, inverse reasoning theory and optimal subset regression, fully excavates useful information hidden in actual load data, effectively solves the problem of failure of adjacent data of a periodic algorithm in a traditional model, and considers the great influence of recent load data on a prediction result and the influence of periodic change of the load data in different time scales.
When the prediction model combining the fuzzy averaging function and the optimal subset regression is applied to short-term power load prediction, influence factors of loads such as weather, date types, economic levels and the like do not need to be considered, and the prediction model has the advantages of simplicity and convenience in implementation, strong operability and good prediction effect, so that the actual application value of the model is high.
Based on the method, the invention also provides equipment for realizing the short-term power load forecasting method, which comprises the following steps: a memory for storing a computer program and a short-term power load prediction method;
a processor for executing the computer program and the short term power load prediction method to realize the steps of the short term power load prediction method.
The present invention also provides a readable storage medium having a short-term power load prediction method based on the above method, the readable storage medium having stored thereon a computer program, the computer program being executed by a processor to implement the steps of the short-term power load prediction method.
In the following, it is assumed that the terminal is a mobile terminal, however, it will be understood by those skilled in the art that the configuration according to the embodiment of the present invention can be applied to a fixed type terminal, in addition to elements particularly used for mobile purposes.
The present invention also provides methods of implementing short-term power load forecasting by combining the exemplary elements and algorithm steps described in connection with the embodiments disclosed herein, which may be embodied in electronic hardware, computer software, or combinations thereof, wherein the exemplary elements and steps have been described above generally in terms of their functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. A method of short-term power load prediction, the method comprising:
configuring a power load data sample sequence;
configuring a power load data sample sequence to a membership function;
combining an improved fuzzy averaging function calculation mode of an inverse reasoning theory with an optimal subset regression algorithm to deduce a power load short-term prediction model;
firstly, defining a continuation sequence of a fuzzy averaging function as a free variable based on an optimal subset regression algorithm;
independent variables are freely combined;
and respectively establishing a linear regression equation by the combined sequences and the dependent variable, and screening a short-term load prediction model of the fuzzy average generation function from all the regression equations according to a preset screening standard to obtain a corresponding load prediction value.
2. The short term power load prediction method according to claim 1,
the step of configuring the power load data to a membership function comprises the following steps:
Figure FDA0003298198130000011
wherein: mu is preset according to the past actual value;
if the time-ordered power load data sample sequence is a periodic sequence, the membership function is:
Figure FDA0003298198130000012
wherein: l represents the period length of the sequence; r is a constant determined empirically or by trial calculation;
if the action of the data in the power load data sample sequence on the prediction point gradually decreases with the distance and shows the periodicity of the power load data, the membership function is as follows:
Figure FDA0003298198130000021
3. the short term power load prediction method according to claim 1,
for a time-ordered power load data sample sequence x (t), defining a fuzzy averaging function according to the following formula:
Figure FDA0003298198130000022
wherein: 1, 2, …, l; l is more than or equal to 1 and less than or equal to m; n islINT (n/l); m ═ INT (n/2) or INT (n/3); l is the period corresponding to the fuzzy averaging function; INT is data rounding.
4. The short term power load prediction method according to claim 1,
the fuzzy averaging function is:
Figure FDA0003298198130000023
wherein: rl＝n-nl·l；RlThe number of the residual items with the number of the samples being n;
will R tolThe + i term is used as the starting point of calculating the fuzzy average generating function and is calculated to the last term according to a preset interval, wherein n-RlIs a multiple of the period l.
5. The short term power load prediction method according to claim 4,
expanding a fuzzy averaging function on a power load data sample sequence to the whole prediction interval, and performing periodic continuation prediction;
the periodic continuation prediction mode is as follows:
Figure FDA0003298198130000031
wherein: 1, 2, …, n; mod represents a numerical remainder; obtaining a periodic continuation matrix of the fuzzy average generation function:
G＝(gi，j)n×l，gi，j≡gl(t) (7)
Figure FDA0003298198130000032
wherein: n x l is the periodic extension momentThe order of the array;
Figure FDA0003298198130000033
is the fuzzy averaging function sequence generated when l is 1;
Figure FDA0003298198130000034
show taking in sequence
Figure FDA0003298198130000035
One of them;
Figure FDA0003298198130000036
show taking in sequence
Figure FDA0003298198130000037
Figure FDA0003298198130000038
One, the rest, and so on.
6. The short term power load prediction method according to claim 1,
screening out a short-term load prediction model of the fuzzy averaging function, and obtaining a corresponding load prediction value in a mode comprising the following steps:
1) performing difference processing on an original sample sequence x (t), and obtaining two difference sequences by means of total calculation;
x(1)(t)＝{Δx(1)，Δx(2)，…，Δx(n-1)} (9)
x(2)(t)＝{Δ2x(1)，Δ2x(2)，…，Δ2x(n-1)} (10)
2) setting mu to be 0.01 based on the membership function;
calculating fuzzy average generation function sequences of three adjacent period power load data sample sequences according to the formula (5), and respectively recording the fuzzy average generation function sequences as
Figure FDA0003298198130000041
And
Figure FDA0003298198130000042
then, the corresponding cycle continuation sequences g are respectively calculated by the formula (6)l (0)(t)、gl (1)(t) and gl (2)(t)；
3) Calculating an accumulated sequence of first order difference sequences:
Figure FDA0003298198130000043
wherein: gl (3)(1)＝x(1)；t＝2，3，…，n；l＝1，2，…，m；
Generating 4m fuzzy average generating function continuation sequences gl (0)(t)、gl (1)(t)、gl (2)(t) and gl (3)(t)；
4) Reference to a two-score criterion pair of computed continuation sequences
gl (0)(t)、gl (1)(t)、gl (2)(t) and gl (3)(t) screening;
the rationale for the double scoring criterion is as follows:
CSC＝M1+M2 (12)
wherein: m1Scoring the quantity; m2 Trend towardsGrading;
the quantitative score is defined as:
Figure FDA0003298198130000051
wherein: r2Is a complex correlation coefficient; n is the length of the sample sequence; qKIs the sum of the squares of the residuals of the prediction model; qXIs the sum of the squares of the total deviations of the prediction model;
and the trend score is according to the minimum discrimination information statistic criterion:
M2＝2[T1+(n-1)·ln(n-1)-T2-T3] (14)
wherein:
Figure FDA0003298198130000052
g is the number of categories of the predicted trend;
when the CSC value is maximum, the corresponding prediction model has the best effect, so the maximum CSC value is used as the standard for screening the optimal subset;
5) respectively carrying out unitary regression calculation on the obtained extension sequence of the fuzzy average generation function and the original power load data sample sequence, and solving the CSC value of the extension sequence;
then chi fang test is carried out to lead the CSC value to be larger than chi2 αThe corresponding sequence of (2) is defined as a prediction factor of a prediction equation, and the number of the prediction factor is recorded as k, namely k free variables are obtained;
according to the Newton's binomial theorem, 2 can be calculatedk-1 permutation combination, each combination to constitute a subset of the load prediction equation;
6) selecting the prediction factors obtained by rough selection;
2 obtained by calculationk-1 subset respectively constructing a multiple linear regression equation of the original power load data sample sequence;
solving the CSC value of the regression result according to the double-scoring criterion;
from 2k1, selecting the subset with the maximum CSC value as the optimal subset of the prediction equation;
7) assuming that the optimal subset contains p independent variables, the obtained short-term load prediction model based on the fuzzy averaging function is as follows:
Figure FDA0003298198130000061
if load prediction of w steps is to be completed, the sequence g is carried outi(t) (i ═ 1, 2, …, p) w steps cycle prolongation is carried out by referring to equation (6), and then the w steps cycle prolongation is substituted into the deduced short-term load prediction model to obtain a corresponding load prediction value。
7. The short term power load prediction method according to claim 1,
and quantitatively evaluating the accuracy of each prediction model, wherein the error evaluation function adopts a Relative Error (RE), a Mean Absolute Percent Error (MAPE) and a Root Mean Square Error (RMSE), and the expressions are respectively as follows:
Figure FDA0003298198130000071
Figure FDA0003298198130000072
Figure FDA0003298198130000073
wherein: n is the number of test samples; x is the number oftThe actual load value at the t-th moment;
Figure FDA0003298198130000074
is the predicted load value at the t-th time.
8. The short term power load prediction method according to claim 1,
the mode for configuring the power load data sample sequence comprises the following steps:
presetting at least three power load data acquisition cycles;
acquiring a plurality of power load data in each power load data acquisition period;
calculating the average value and the standard deviation of the power load data in the power load data acquisition period;
setting a power load data comparison parameter range, wherein the power load data comparison parameter range is the average value of power load data plus or minus 2 multiplied by the standard deviation of the power load data;
and comparing each power load data in a preset time period with the power load data comparison parameter range, if the power load data comparison parameter range is exceeded, deleting the current power load data, and calling the power load data from the subsequent period of the power load data acquisition period for supplement.
9. An apparatus for implementing a short-term power load forecasting method, comprising:
a memory for storing a computer program and a short-term power load prediction method;
a processor for executing the computer program and the short term power load prediction method to implement the steps of the short term power load prediction method as claimed in any one of claims 1 to 8.
10. A readable storage medium having a short term power load prediction method, wherein the readable storage medium has stored thereon a computer program which is executed by a processor to implement the steps of the short term power load prediction method as claimed in any one of claims 1 to 8.
CN202111183431.2A 2021-10-11 2021-10-11 Short-term power load prediction method and device and readable storage medium Pending CN113822492A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111183431.2A CN113822492A (en) 2021-10-11 2021-10-11 Short-term power load prediction method and device and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111183431.2A CN113822492A (en) 2021-10-11 2021-10-11 Short-term power load prediction method and device and readable storage medium

Publications (1)

Publication Number Publication Date
CN113822492A true CN113822492A (en) 2021-12-21

Family

ID=78916358

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111183431.2A Pending CN113822492A (en) 2021-10-11 2021-10-11 Short-term power load prediction method and device and readable storage medium

Country Status (1)

Country Link
CN (1) CN113822492A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116562171A (en) * 2023-07-07 2023-08-08 南京市计量监督检测院 Error assessment method for online measurement of temperature and humidity
CN116992274A (en) * 2023-09-28 2023-11-03 国网山东省电力公司滨州市滨城区供电公司 Short-term wind speed prediction method and system based on improved principal component regression model

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105809286A (en) * 2016-03-08 2016-07-27 南昌工程学院 Representative data reconstruction-based incremental SVR (support vector regression) load prediction method
CN110611333A (en) * 2019-11-11 2019-12-24 燕山大学 Island microgrid distributed coordination control method considering communication problem
CN111131063A (en) * 2019-12-31 2020-05-08 腾讯科技（深圳）有限公司 Flow control method, related device, equipment and storage medium
CN112712203A (en) * 2020-12-29 2021-04-27 湖南大学 Method and system for predicting daily maximum load of power distribution network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105809286A (en) * 2016-03-08 2016-07-27 南昌工程学院 Representative data reconstruction-based incremental SVR (support vector regression) load prediction method
CN110611333A (en) * 2019-11-11 2019-12-24 燕山大学 Island microgrid distributed coordination control method considering communication problem
CN111131063A (en) * 2019-12-31 2020-05-08 腾讯科技（深圳）有限公司 Flow control method, related device, equipment and storage medium
CN112712203A (en) * 2020-12-29 2021-04-27 湖南大学 Method and system for predicting daily maximum load of power distribution network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张健: "异常电力负荷数据的t检验辨识与修正", 《电力需求侧管理》 *
王为国等: "基于逆推理论改进模糊均生函数的短期负荷预测", 《水电能源科学》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116562171A (en) * 2023-07-07 2023-08-08 南京市计量监督检测院 Error assessment method for online measurement of temperature and humidity
CN116562171B (en) * 2023-07-07 2023-09-12 南京市计量监督检测院 Error assessment method for online measurement of temperature and humidity
CN116992274A (en) * 2023-09-28 2023-11-03 国网山东省电力公司滨州市滨城区供电公司 Short-term wind speed prediction method and system based on improved principal component regression model
CN116992274B (en) * 2023-09-28 2024-02-02 国网山东省电力公司滨州市滨城区供电公司 Short-term wind speed prediction method and system based on improved principal component regression model

Similar Documents

Publication Publication Date Title
Liu et al. Single-machine-based joint optimization of predictive maintenance planning and production scheduling
CN113822492A (en) Short-term power load prediction method and device and readable storage medium
Wan et al. Price-based residential demand response management in smart grids: A reinforcement learning-based approach
Zhou et al. Energy-efficient scheduling of a single batch processing machine with dynamic job arrival times
CN115600774B (en) Multi-target production scheduling optimization method for assembly type building component production line
CN112633316A (en) Load prediction method and device based on boundary estimation theory
CN112232600A (en) Short-term power load prediction method and system based on variational modal decomposition
Rokhforoz et al. Multi-agent maintenance scheduling based on the coordination between central operator and decentralized producers in an electricity market
CN112446537A (en) Short-term load prediction method based on deep long-term and short-term memory network
CN111222787A (en) Receiving-end power grid large-scale energy storage business mode decision method and system
Ye et al. Competing failure modeling for performance analysis of automated manufacturing systems with serial structures and imperfect quality inspection
Liu et al. Multi-component manufacturing system maintenance scheduling based on degradation information using genetic algorithm
Mobin et al. Multi-objective X-bar control chart design by integrating NSGA-II and data envelopment analysis
Li et al. An expert knowledge-based dynamic maintenance task assignment model using discrete stress–strength interference theory
CN110837934B (en) Smart grid short-term residential load prediction method based on deep learning
Al-Aomar A robust simulation-based multicriteria optimization methodology
Cano et al. MRP systems considering fuzzy capacity, lead times and inventory availability
Liu et al. Kernel estimation of quantile sensitivities
CN112508260B (en) Medium-and-long-term load prediction method and device of distribution transformer based on comparative learning
CN112288278A (en) Equipment asset management method based on knowledge graph
CN112183872A (en) Blast furnace gas generation amount prediction method combining generation of countermeasure network and neural network
CN111628498A (en) Multi-target power distribution network reconstruction method and device considering power distribution network reliability
Alonso et al. Measuring the pro-activity of software agents
CN115629576A (en) Non-invasive flexible load aggregation characteristic identification and optimization method, device and equipment
CN114723277A (en) Method, apparatus, device, medium, and program product for performance evaluation

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: 20211221