CN111695739A - Load prediction method, system and equipment - Google Patents
Load prediction method, system and equipment Download PDFInfo
- Publication number
- CN111695739A CN111695739A CN202010554471.2A CN202010554471A CN111695739A CN 111695739 A CN111695739 A CN 111695739A CN 202010554471 A CN202010554471 A CN 202010554471A CN 111695739 A CN111695739 A CN 111695739A
- Authority
- CN
- China
- Prior art keywords
- mapping table
- factor mapping
- relevant
- related factor
- function
- 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.)
- Granted
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/10—Power transmission or distribution systems management focussing at grid-level, e.g. load flow analysis, node profile computation, meshed network optimisation, active network management or spinning reserve management
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- 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
Abstract
The invention discloses a load prediction method, a system and equipment, comprising the following steps: constructing a relevant factor mapping table, and constructing a mapping function according to the relevant factor mapping table; training the mapping function based on a perturbation method to obtain a trained first relevant factor mapping table; training the relevant factor mapping table based on a genetic algorithm to obtain a second trained relevant factor mapping table; the method adopts the perturbation method and the genetic algorithm to train the mapping function and the related factor mapping table respectively, improves the accuracy of the related factor mapping table, adopts the prediction decision algorithm to select the final related factor mapping table with smaller error, further improves the accuracy of load prediction, and solves the technical problem that the related factor mapping table designed according to experience in the prior art has lower prediction accuracy when the load is predicted.
Description
Technical Field
The present invention relates to the field of load prediction, and in particular, to a load prediction method, system and device.
Background
At present, in the traditional short-term load prediction, the processing of the value of the relevant factor has a great influence on the final prediction precision, and the load is generally predicted by adopting a method of a relevant factor mapping table in the past. However, the mapping table length of each relevant factor is defined by the user (generally by virtue of customer experience), and each mapping pair (the value of the relevant factor < - > after mapping) is also the result of experience earned by the user. A user designs a plurality of mapping table schemes for virtual prediction, and selects one with satisfactory effect as a mapping table of the region of the user. However, the accuracy of the finally obtained load predicted value is low due to subjective initiative in the process of designing the mapping table scheme.
In summary, the related factor mapping table designed according to experience in the prior art has a technical problem of low prediction accuracy when predicting the load.
Disclosure of Invention
The invention provides a load prediction method, a system and equipment, which are used for solving the technical problem that the prediction accuracy is low when a load is predicted by a related factor mapping table designed according to experience in the prior art.
The invention provides a load prediction method, which comprises the following steps:
s1: acquiring related factor data to construct a related factor mapping table, and constructing a mapping function according to the related factor mapping table;
s2: training the mapping function based on a perturbation method to obtain a trained mapping function; applying the trained mapping function to a related factor mapping table to obtain a trained first related factor mapping table;
s3: training the relevant factor mapping table based on a genetic algorithm to obtain a second trained relevant factor mapping table;
s4: and selecting a final related factor mapping table from the first related factor mapping table and the second related factor mapping table by adopting a prediction decision algorithm, and predicting the load according to the final related factor mapping table.
Preferably, the specific process of training the mapping function based on the perturbation method is as follows:
s201: setting a precision threshold value and iteration times, and acquiring a related factor B from a related factor mapping table(k)Wherein
S202: let k equal to 0, correlate factor B(0)Inputting the prediction result into a mapping function, and calculating the prediction precision f of the mapping functionopt;
S203: making a correlation factor counter i equal to 1, and making an ith correlation factor mapping table segmentation counter j equal to 1;
s205: let B ═ B(k)+ΔB′、B″=B(k)+ Δ B ", inputting B 'and B" into the mapping function respectively, and solving the prediction precisions f' and f "of the mapping function;
s206: f ', f' and foptF as new when the prediction accuracy is highestoptNew foptThe corresponding related factor is B(k+1);
S207: j is judged>=liIf yes, executing step S209, otherwise, making j equal to j +1, and executing step S204 again;
s208: judging whether i > -m is true or not; if yes, let K be K +1, execute S2010; if not, let j equal j +1, go to S203;
s209: judgment of foptWhether the current mapping function is smaller than the precision threshold value or not is judged, and if the current mapping function is smaller than the precision threshold value, the current mapping function is output; if not, executing S2010;
s2010: judging whether k is larger than the iteration times, if so, outputting the current mapping function; if not, go to S202.
Preferably, the specific process of training the relevant factor mapping table based on the genetic algorithm is as follows:
s301: determining parameters and an evaluation function of a genetic algorithm;
s302: selecting relevant factors from the relevant factor mapping table, coding the relevant factors to form an initial chromosome, and inputting the initial chromosome into a genetic algorithm;
s303: carrying out cross generation on the initial chromosome in a genetic algorithm to generate a progeny chromosome, and carrying out mutation on the progeny chromosome;
s304: calculating the fitness of all chromosomes by using an evaluation function, judging whether the chromosomes accord with the optimization criterion according to the fitness, and if so, outputting the best chromosome and decoding to obtain the best solution; otherwise, the offspring chromosome is taken as the initial chromosome, and step S303 is executed again.
Preferably, the specific process of training the relevant factor mapping table based on the genetic algorithm is as follows:
and (4) coding the relevant factors by adopting a real number coding mode to form an initial chromosome.
Preferably, the specific process for determining the parameters of the genetic algorithm is as follows:
the crossover probability, mutation probability, and number of initial chromosomes of the genetic algorithm are determined.
Preferably, the specific process of selecting the final related factor mapping table from the first related factor mapping table and the second related factor mapping table by using a prediction decision algorithm is as follows:
s401: evaluating the first relevant factor mapping table and the second relevant factor mapping table by adopting a probability matrix method, and calculating the weight of the first relevant factor mapping table and the weight of the second relevant factor mapping table according to the evaluation result;
s402: calculating a probability distribution function of the first related factor mapping table weight and a probability distribution function of the second related factor mapping table weight, and respectively calculating a mathematical expected value of the probability distribution function of the first related factor mapping table weight and a mathematical expected value of the probability distribution function of the second related factor mapping table weight;
s403: and selecting a final related factor mapping table from the first related factor mapping table and the second related factor mapping table according to the mathematical expected value.
Preferably, the weight of the first correlation factor mapping table and the weight of the second correlation factor mapping table are calculated by using a least square method or a feature vector method.
Preferably, the specific process of calculating the probability distribution function of the first related factor mapping table weight and the probability distribution function of the second related factor mapping table weight is as follows:
and (3) dividing the interval [0,1] into N equal intervals, counting the frequency of values of the first relevant factor mapping table weight and the second relevant factor mapping table weight in each equal interval, and dividing the frequency of the values by two to obtain the distribution probability function of the first relevant factor mapping table weight and the second relevant factor mapping table weight in the interval [0,1 ].
A load prediction system comprises a related factor mapping table construction module, a perturbation method module, a genetic algorithm module and a prediction decision algorithm module;
the relevant factor mapping table construction module is used for acquiring relevant factor data and constructing a relevant factor mapping table;
the perturbation method module is used for constructing a mapping function according to the relevant factor mapping table, and training the mapping function based on the perturbation method to obtain a trained first relevant factor mapping table;
the genetic algorithm module is used for training the relevant factor mapping table based on a genetic algorithm to obtain a trained second relevant factor mapping function;
and the prediction decision algorithm module is used for selecting a final related factor mapping table from the first related factor mapping table and the second related factor mapping table by adopting a prediction decision algorithm and predicting the load according to the final related factor mapping table.
A load prediction device comprising a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to perform one of the load prediction methods described above according to instructions in the program code.
According to the technical scheme, the embodiment of the invention has the following advantages:
the embodiment of the invention extracts the mapping function in the relevant factor mapping table by constructing the relevant factor mapping table; and training the mapping function and the related factor mapping table respectively by adopting a perturbation method and a genetic algorithm to obtain a trained first related factor mapping table and a trained second related factor mapping table, so that the accuracy of the related factor mapping tables is improved, finally, a prediction decision algorithm is adopted to select a final related factor mapping table with a smaller error from the first related factor mapping table and the second related factor mapping table, and the load is predicted according to the final related factor mapping table, so that the accuracy of the related factor mapping value is further improved, and the technical problem that the prediction accuracy is lower when the load is predicted due to the fact that the related factor mapping table is designed according to experience in the prior art is solved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without inventive exercise.
Fig. 1 is a flowchart of a method, a system and a device for load prediction according to an embodiment of the present invention.
Fig. 2 is a monotonic correlation factor mapping graph of a load prediction method, system and device according to an embodiment of the present invention.
Fig. 3 is a non-monotonic correlation factor mapping graph of a load prediction method, system and device according to an embodiment of the present invention.
Fig. 4 is a flowchart of training a mapping function based on a perturbation method in a load prediction method, system and device according to an embodiment of the present invention.
Fig. 5 is a flowchart of a load prediction method, system, and device provided in the embodiments of the present invention for training a related factor mapping table based on a genetic algorithm.
Fig. 6 is a system framework diagram of a method, a system, and a device for load prediction according to an embodiment of the present invention.
Fig. 7 is a device framework diagram of a method, a system, and a device for load prediction according to an embodiment of the present invention.
Detailed Description
The embodiment of the invention provides a load prediction method, a system and equipment, which are used for solving the technical problem that the prediction accuracy is low when a load is predicted by a related factor mapping table designed according to experience in the prior art.
In order to make the objects, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the embodiments described below 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.
Referring to fig. 1, fig. 1 is a flowchart illustrating a method, a system and a device for predicting power consumption of a user according to an embodiment of the present invention.
The load prediction method provided by the embodiment of the invention comprises the following steps:
s1: acquiring relevant factor data from a background of the power system, constructing a relevant factor mapping table according to the relevant factor data, and constructing a mapping function according to the relevant factor mapping table;
it should be further explained that the process of constructing the relevant factor mapping table is as follows: first, several variables are defined as follows: the number of correlation factors is m, and the number of data sampling days is N + N days, where 1 → N days correlation factors and load values are known. And (3) only relevant factors are known for days N +1 → N + N, and the load value is to be predicted. The index of the relevant factor is i; day subscript: k; time period subscript: t, liThe number of the segmented nodes of the ith correlation factor mapping table is T, and the number of the load points per day is T;
the mapping table for the ith relevant factor is described as follows:
wherein the content of the first and second substances,an argument representing the mapping table is used,a dependent variable representing a mapping table;
the calculation relationship between the known correlation factor data and the load value to be predicted is shown in table 1:
TABLE 1
Wherein x iskiRepresenting the value before mapping of the ith relevant factor on the kth day; y iskiExpressing a mapped value obtained by searching a mapping table for the ith relevant factor on the kth day; lktActual values representing the load at point t on day k;the predicted value of the load at the t-th point on the k-th day is shown.
Wherein, the value vector is taken after mappingEach element b ofijAll of them are required to be greater than 0 and generally lie in the interval [0,1]]In, but not necessarily monotonous toNamely: with aijIs an independent variable, is bijThere is a possibility of inflection points appearing in the curve drawn for the dependent variable, and as shown in fig. 2 and 3, the mapping process of fig. 2 and 3 is:
(1) for the normal point: directly obtaining a mapping value;
(2) for intermediate points (no direct mapping values): performing interpolation, such as linear interpolation, to obtain a mapping value;
(3) for out-of-endpoint: and performing level ratio extension or linear extension to obtain a mapping value.
Hereinafter, day 1 → N is referred to as set 1, and day N +1 → N + N is referred to as set 2, there are:
X=[X1,X2]for the correlation factor original value matrix, X is ═ X1,X2]After the elements in the data are mapped, a related factor mapping value matrix can be obtained;
y ═ g (X, a, B) is a functional abstraction of the mapping process, and Y1=g(X1,A,B),Y2=g(X2,A,B);
L1=h(Y1S) is a functional abstraction of the fitting process for learning the intrinsic parameters S of the mapping table.
Is an abstraction of the virtual prediction process and is used for predicting mapping values according to the intrinsic parameters S of the mapping table.
In summary, the following results can be obtained: and L ═ f (X, a, B, S) is the mapping function. The meaning is as follows: after the parameters A and B of the mapping library are given, the predicted value can be determined after the value of the known relevant factor X is taken. And a group of values (A and B) is changed, and the predicted value is likely to change under the condition that X is not changed. Then, when (a, B) is constantly changing, the predicted value may be optimized.
S2: training the mapping function based on perturbation method, and for the condition of only adjusting the mapping value of the relevant factor, the variable (decision variable) to be optimized is the dependent variable of the mapping tableThe perturbation method has the basic idea that each component of B generates positive perturbation and negative perturbation in each iteration, then the prediction effect of the mapping function after perturbation is judged, the best prediction effect is taken as the next solution vector in the positive perturbation, the non-perturbation and the negative perturbation, the dependent variable B of the mapping table is iterated continuously until the prediction effect of the mapping function meets the requirement, so that a trained mapping function is obtained, and the trained mapping function is applied to a related factor mapping table to obtain a trained first related factor mapping table;
s3: training a relevant factor mapping table based on a genetic algorithm, wherein the genetic algorithm is a method for searching an optimal solution by simulating a natural evolution process; in the embodiment, the related factors are input into a genetic algorithm as chromosomes, the solving process of the mapping values of the related factors is converted into the processes of crossing, variation and the like of chromosome genes in the similar biological evolution, and the optimal solution is obtained through an adaptive function, so that a trained second related factor mapping table is obtained;
s4: and selecting a final related factor mapping table from the first related factor mapping table and the second related factor mapping table by adopting a prediction decision algorithm, wherein the target of the prediction decision algorithm is to select one with more remarkable prediction precision from the first related factor mapping table and the second related factor mapping table as a prediction result, while the other one is abandoned, and finally, the load is predicted by obtaining the related factor mapping value according to the final related factor mapping table, so that the accuracy of load prediction is further improved.
As a preferred embodiment, as shown in fig. 4, the specific process of training the mapping function based on perturbation method is:
s201: setting a precision threshold and iteration times, wherein the iteration times are set to be 1000 times according to an empirical value, so that a certain amount of iteration times are ensured, and the consumed time is not too long; obtaining relevant factor B from relevant factor mapping table(k)Wherein
S202: let k equal to 0, correlate factor B(0)Inputting the prediction result into a mapping function, and calculating the prediction precision f of the mapping functionopt(ii) a The algorithm for obtaining the prediction precision can be randomly selected, and for short-term load prediction, prediction can be carried out through one of a regression algorithm, multiple proportion smoothing, an artificial neural network, point-to-point prediction, Adaboost, deep learning and the like;
s203: making a correlation factor counter i equal to 1, and making an ith correlation factor mapping table segmentation counter j equal to 1;
s204: suppose thatWherein a and β are weights of related factors, and Δ B' are respectively B(0)Positive and negative perturbation components of;
s205: let B ═ B(k)+ΔB′、B″=B(k)+ Δ B ", inputting B 'and B" into the mapping function respectively, and solving the prediction precisions f' and f "of the mapping function;
s206: f ', f' and foptF as new when the prediction accuracy is highestoptNew foptThe corresponding related factor is B(k+1)So that the solution vector with the best prediction effect is taken as the next solution vector in the positive perturbation, the non-perturbation and the negative perturbation;
s207: j is judged>=liIf yes, executing step S209, otherwise, making j equal to j +1, and executing step S204 again;
s208: judging whether i > -m is true or not; if yes, let K be K +1, execute S2010; if not, let j equal j +1, go to S203;
s209: judgment of foptWhether the current mapping function is smaller than the precision threshold value or not is judged, and if the current mapping function is smaller than the precision threshold value, the current mapping function is output; if not, executing S2010;
s2010: when iteration reaches a certain count, the prediction precision of the mapping function can reach the optimum, and whether k is greater than the iteration times is judged; if yes, outputting the current mapping function; if not, go to S202.
It should be noted that the iteration number is set to 1000 according to an empirical value, so that a certain number of iteration numbers is ensured, and the time consumption is not too long.
As a preferred embodiment, as shown in fig. 5, the specific process of training the correlation factor mapping table based on the genetic algorithm is as follows:
s301: determining parameters of a genetic algorithm and an evaluation function, wherein the parameters of the genetic algorithm comprise cross probability, mutation probability and the number of initial chromosomes; the cross probability is used for judging whether two chromosomes need to be crossed; the mutation probability is used for judging whether any chromosome needs mutation;
s302: selecting relevant factors from the relevant factor mapping table, coding the relevant factors by adopting a real number coding mode to form an initial chromosome, and inputting the initial chromosome into a genetic algorithm;
s303: crossing the initial chromosomes based on the crossing probability to generate offspring chromosomes, and carrying out mutation on the offspring chromosomes based on the mutation probability to obtain mutated offspring chromosomes;
s304: calculating the fitness of all chromosomes by using an evaluation function, judging whether the chromosomes accord with the optimization criterion according to the fitness, and if so, outputting the best chromosome and decoding to obtain the best solution; otherwise, the offspring chromosome is taken as the initial chromosome, and step S303 is executed again. The results of the correlation factor mapping table after being trained by the genetic algorithm are shown in table 2.
And selecting the average deviation ratio of the load and the real historical load as an optimized objective function. The optimization criterion is to optimize the direction with small average deviation ratio of the load and the real historical load. The virtual prediction precision is used as a basis for evaluating the chromosome quality, and the evaluation process is essentially a one-time virtual prediction process. Firstly, calling various single prediction methods to virtually predict a certain historical date, then calling an integrated model according to weights of different methods to virtually predict the same date, respectively predicting the electric quantity of a time sequence to be predicted by the integrated model through several different prediction algorithms, then multiplying each prediction result by a weight, adding the results to obtain a final prediction result, and the integrated model can integrate the advantages of the prediction methods, can effectively improve the precision to finally obtain the virtual prediction load of the date, and judges the quality of a chromosome by comparing the virtual prediction load with the historical load.
Wherein, PdtRepresents: actual value of load at point t on day d;the predicted value of the load at the t-th point on the day d is shown. I is the virtual prediction days, and T represents the sampling point number of the electric quantity.
TABLE 2
As a preferred embodiment, the specific process of selecting the final related factor mapping table from the first related factor mapping table and the second related factor mapping table by using the prediction decision algorithm is as follows:
s401: evaluating the first relevant factor mapping table and the second relevant factor mapping table by adopting a probability matrix method, and calculating the weight of the first relevant factor mapping table and the weight of the second relevant factor mapping table according to the evaluation result;
it should be further explained that the probability matrix method is a method for solving the multi-objective optimization problem in the decision theory, and the principle thereof is as follows:
assuming q prediction methods, the practical advantages and disadvantages of the q prediction methods are represented by a weight vector W, where W is { W ═ W1,w2,...,wq}TWeight wkLarger indicates better process. Defining a decision matrix O, and introducing a weight ratio matrix R, then:
order:
O≈R
then
Okj≈wk/wj,k,j∈{1,2,...,q}
Each element O in the decision matrix OkjOne can see the probability that method k outperforms method j, k, j ∈ {1, 2.., q }. where each element in the decision matrix O is a positive number, and O is a positive numberkk=1。
If the decision maker is right to OkjAre consistent with the estimation of
Okj=1/Ojk,k,j∈{1,2,...,q}
Okj=OkiOij,k,j,i∈{1,2,...,q}
The method of estimating the decision matrix O according to the concept of probability (i.e. the ratio of the probability of a random event to the probability of its non-occurrence) is as follows:
1. let PikjRepresents the probability that method k outperforms method j in the next actual prediction;
2. by a ratio of pikj/πjkRepresenting the probability of method k being superior to method j, i.e. Okj=πkj/πjk;
When the prediction experiment was performed, it was assumed that n points in history were predicted using method k and method j, respectively. a iskjRepresenting the number of times method k outperformed method j, ajkRepresenting the number of times (a) method j outperformed method kkj+ajkN). Then the ratio akjN represents pikj(ii) a Ratio ajkN represents pijk。
The first relevant factor mapping table obtained by the perturbation method and the second relevant factor mapping table obtained by the genetic algorithm are evaluated by the method, so that the probability that the perturbation method is superior to the genetic algorithm is obtained.
If the decision maker is right to OkjIf the estimates of (are) not consistent, then there is
Okj≈wk/wj
Is superior to Okjwj-wkIs not 0, so a set of weights w is selected1,w2,...,wqMinimize the sum of the squares of the errors, i.e.
Weight { w ] in the above equation1,w2,...,wqIs constrained to
This is a typical quadratic programming problem, and the weights of each prediction method can be obtained by applying a standard algorithm to solve.
S402: calculating a probability distribution function of the first related factor mapping table weight and a probability distribution function of the second related factor mapping table weight, and respectively calculating a mathematical expectation value of the probability distribution function of the first related factor mapping table weight and a mathematical expectation value of the probability distribution function of the second related factor mapping table weight, wherein the specific process comprises the following steps:
obtaining a weight vector W ═ W for each prediction method1,w2,...,wqAfter f due to wi∈[0,1]Thus, the interval [0,1]]Divided into K equal parts (each interval having a width of) Counting the frequency of the values of the weight of each prediction method in each equally divided interval and dividing the frequency by the total number q of the prediction methods to obtain the interval [0,1] of the weight]Approximate distribution probability of (p) { p ═ p }1,...,pKIn which p (x)1)=p1,…,p(xK)=pK,{x1,...,xKThe mean value of the start and stop values for each equal interval is considered:
……
the mathematical expectation of the weights is
EV=p1x1+p2x2+...+pKxK。
S403: selecting a related factor mapping table with the weight probability above a mathematical expectation value from the first related factor mapping table and the second related factor mapping table as a final related factor mapping table;
the values of the correlation factor mapping values of the final correlation factor mapping table are shown in table 3.
TABLE 3
As shown in fig. 6, a load prediction system includes a related factor mapping table constructing module 201, a perturbation method module 202, a genetic algorithm module 203, and a prediction decision algorithm module 204;
the related factor mapping table constructing module 201 is configured to obtain related factor data and construct a related factor mapping table;
the perturbation method module 202 is configured to construct a mapping function according to the related factor mapping table, and train the mapping function based on the perturbation method to obtain a trained first related factor mapping table;
the genetic algorithm module 203 is used for training the relevant factor mapping table based on a genetic algorithm to obtain a trained second relevant factor mapping function;
the prediction decision algorithm module 204 is configured to select a final related factor mapping table from the first related factor mapping table and the second related factor mapping table by using a prediction decision algorithm, and obtain a related factor mapping value according to the final related factor mapping table.
As shown in fig. 7, an electric device load data monitoring and analyzing device 30 includes a processor 300 and a memory 301;
the memory 301 is used for storing a program code 302 and transmitting the program code 302 to the processor;
the processor 300 is configured to execute the steps of the above-mentioned method for monitoring and analyzing load data of electric equipment according to the instructions in the program code 302.
Illustratively, the computer program 302 may be partitioned into one or more modules/units that are stored in the memory 301 and executed by the processor 300 to accomplish the present application. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution process of the computer program 302 in the terminal device 30.
The terminal device 30 may be a desktop computer, a notebook, a palm computer, a cloud server, or other computing devices. The terminal device may include, but is not limited to, a processor 300, a memory 301. Those skilled in the art will appreciate that fig. 7 is merely an example of a terminal device 30, and does not constitute a limitation of terminal device 30, and may include more or fewer components than shown, or some components in combination, or different components, e.g., the terminal device may also include input-output devices, network access devices, buses, etc.
The Processor 300 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf ProgrammaBle Gate Array (FPGA) or other ProgrammaBle logic device, discrete Gate or transistor logic, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The storage 301 may be an internal storage unit of the terminal device 30, such as a hard disk or a memory of the terminal device 30. The memory 301 may also be an external storage device of the terminal device 30, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a flash memory Card (FlashCard), and the like, which are provided on the terminal device 30. Further, the memory 301 may also include both an internal storage unit and an external storage device of the terminal device 30. The memory 301 is used for storing the computer program and other programs and data required by the terminal device. The memory 301 may also be used to temporarily store data that has been output or is to be output.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described device embodiments are merely illustrative, and for example, the division of the units is only one logical functional 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 be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method of load prediction, comprising the steps of:
s1: acquiring related factor data to construct a related factor mapping table, and constructing a mapping function according to the related factor mapping table;
s2: training the mapping function based on a perturbation method to obtain a trained mapping function; applying the trained mapping function to a related factor mapping table to obtain a trained first related factor mapping table;
s3: training the relevant factor mapping table based on a genetic algorithm to obtain a second trained relevant factor mapping table;
s4: and selecting a final related factor mapping table from the first related factor mapping table and the second related factor mapping table by adopting a prediction decision algorithm, and predicting the load according to the final related factor mapping table.
2. The method of claim 1, wherein the training of the mapping function based on perturbation method comprises:
s201: setting a precision threshold value and iteration times, and acquiring a related factor B from a related factor mapping table(k)Wherein
S202: let k equal to 0, correlate factor B(0)Inputting the prediction result into a mapping function, and calculating the prediction precision f of the mapping functionopt;
S203: making a correlation factor counter i equal to 1, and making an ith correlation factor mapping table segmentation counter j equal to 1;
s205: let B ═ B(k)+ΔB′、B″=B(k)+ Δ B ", inputting B 'and B" into the mapping function respectively, and solving the prediction precisions f' and f "of the mapping function;
s206: f ', f' and foptF as new when the prediction accuracy is highestoptNew foptThe corresponding related factor is B(k +1);
S207: j is judged>=liIf yes, executing step S209, otherwise, making j equal to j +1, and executing step S204 again;
s208: judging whether i > -m is true or not; if yes, let K be K +1, execute S2010; if not, let j equal j +1, go to S203;
s209: judgment of foptWhether the current mapping function is smaller than the precision threshold value or not is judged, and if the current mapping function is smaller than the precision threshold value, the current mapping function is output; if not, executing S2010;
s2010: judging whether k is larger than the iteration times, if so, outputting the current mapping function; if not, go to S202.
3. The load prediction method according to claim 1, wherein the specific process of training the correlation factor mapping table based on the genetic algorithm is as follows:
s301: determining parameters and an evaluation function of a genetic algorithm;
s302: selecting relevant factors from the relevant factor mapping table, coding the relevant factors to form an initial chromosome, and inputting the initial chromosome into a genetic algorithm;
s303: carrying out cross generation on the initial chromosome in a genetic algorithm to generate a progeny chromosome, and carrying out mutation on the progeny chromosome;
s304: calculating the fitness of all chromosomes by using an evaluation function, judging whether the chromosomes accord with the optimization criterion according to the fitness, and if so, outputting the best chromosome and decoding to obtain the best solution; otherwise, the offspring chromosome is taken as the initial chromosome, and step S303 is executed again.
4. The load prediction method according to claim 3, wherein the specific process of training the correlation factor mapping table based on the genetic algorithm is as follows:
and (4) coding the relevant factors by adopting a real number coding mode to form an initial chromosome.
5. A method for load prediction according to claim 3, characterized in that the specific process of determining the parameters of the genetic algorithm is:
the crossover probability, mutation probability, and number of initial chromosomes of the genetic algorithm are determined.
6. The load prediction method according to claim 1, wherein the specific process of selecting the final correlation factor mapping table from the first correlation factor mapping table and the second correlation factor mapping table by using a prediction decision algorithm is as follows:
s401: evaluating the first relevant factor mapping table and the second relevant factor mapping table by adopting a probability matrix method, and calculating the weight of the first relevant factor mapping table and the weight of the second relevant factor mapping table according to the evaluation result;
s402: calculating a probability distribution function of the first related factor mapping table weight and a probability distribution function of the second related factor mapping table weight, and respectively calculating a mathematical expected value of the probability distribution function of the first related factor mapping table weight and a mathematical expected value of the probability distribution function of the second related factor mapping table weight;
s403: and selecting a final related factor mapping table from the first related factor mapping table and the second related factor mapping table according to the mathematical expected value.
7. The load prediction method according to claim 6, wherein the first correlation factor mapping table weight and the second correlation factor mapping table weight are calculated by a least square method or a feature vector method.
8. The load prediction method according to claim 6, wherein the specific process of calculating the probability distribution function of the first related factor mapping table weight and the probability distribution function of the second related factor mapping table weight is as follows:
and (3) dividing the interval [0,1] into N equal intervals, counting the frequency of values of the first relevant factor mapping table weight and the second relevant factor mapping table weight in each equal interval, and dividing the frequency of the values by two to obtain the distribution probability function of the first relevant factor mapping table weight and the second relevant factor mapping table weight in the interval [0,1 ].
9. A load prediction system is characterized by comprising a related factor mapping table construction module, a perturbation method module, a genetic algorithm module and a prediction decision algorithm module;
the relevant factor mapping table construction module is used for acquiring relevant factor data and constructing a relevant factor mapping table;
the perturbation method module is used for constructing a mapping function according to the relevant factor mapping table, and training the mapping function based on the perturbation method to obtain a trained first relevant factor mapping table;
the genetic algorithm module is used for training the relevant factor mapping table based on a genetic algorithm to obtain a trained second relevant factor mapping function;
and the prediction decision algorithm module is used for selecting a final related factor mapping table from the first related factor mapping table and the second related factor mapping table by adopting a prediction decision algorithm and predicting the load according to the final related factor mapping table.
10. A load prediction device comprising a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to execute a load prediction method according to any one of claims 1 to 8 according to instructions in the program code.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010554471.2A CN111695739B (en) | 2020-06-17 | 2020-06-17 | Load prediction method, system and equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010554471.2A CN111695739B (en) | 2020-06-17 | 2020-06-17 | Load prediction method, system and equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111695739A true CN111695739A (en) | 2020-09-22 |
CN111695739B CN111695739B (en) | 2022-06-14 |
Family
ID=72481676
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010554471.2A Active CN111695739B (en) | 2020-06-17 | 2020-06-17 | Load prediction method, system and equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111695739B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170091670A1 (en) * | 2015-09-29 | 2017-03-30 | Yandex Europe Ag | Method of and system for generating a prediction model and determining an accuracy of a prediction model |
CN107016469A (en) * | 2017-04-13 | 2017-08-04 | 重庆大学 | Methods of electric load forecasting |
CN107909220A (en) * | 2017-12-08 | 2018-04-13 | 天津天大求实电力新技术股份有限公司 | Electric heating load prediction method |
CN108985570A (en) * | 2018-08-17 | 2018-12-11 | 深圳供电局有限公司 | A kind of load forecasting method and its system |
-
2020
- 2020-06-17 CN CN202010554471.2A patent/CN111695739B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170091670A1 (en) * | 2015-09-29 | 2017-03-30 | Yandex Europe Ag | Method of and system for generating a prediction model and determining an accuracy of a prediction model |
CN107016469A (en) * | 2017-04-13 | 2017-08-04 | 重庆大学 | Methods of electric load forecasting |
CN107909220A (en) * | 2017-12-08 | 2018-04-13 | 天津天大求实电力新技术股份有限公司 | Electric heating load prediction method |
CN108985570A (en) * | 2018-08-17 | 2018-12-11 | 深圳供电局有限公司 | A kind of load forecasting method and its system |
Non-Patent Citations (3)
Title |
---|
严凤 等: "《基于预测决策的组合短期负荷预测》", 《电测与仪表》 * |
高峰 等: "《短期负荷预测相关因素的自适应训练》", 《电力系统自动化》 * |
高峰 等: "《负荷预测中多模型的自动筛选方法》", 《电力系统自动化》 * |
Also Published As
Publication number | Publication date |
---|---|
CN111695739B (en) | 2022-06-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Jiang et al. | Holt–Winters smoothing enhanced by fruit fly optimization algorithm to forecast monthly electricity consumption | |
Bremnes | Ensemble postprocessing using quantile function regression based on neural networks and Bernstein polynomials | |
Yu et al. | Markov decision processes with arbitrary reward processes | |
Ak et al. | An interval-valued neural network approach for uncertainty quantification in short-term wind speed prediction | |
Van Calster et al. | ProfARIMA: A profit-driven order identification algorithm for ARIMA models in sales forecasting | |
Defourny et al. | Scenario trees and policy selection for multistage stochastic programming using machine learning | |
Štula et al. | Continuously self-adjusting fuzzy cognitive map with semi-autonomous concepts | |
Rivero et al. | Energy associated tuning method for short-term series forecasting by complete and incomplete datasets | |
CN114330863A (en) | Time series prediction processing method, device, storage medium and electronic device | |
CN111027672A (en) | Time sequence prediction method based on interactive multi-scale recurrent neural network | |
CN110264007A (en) | Stock exchange control method and device | |
CN114817571B (en) | Dynamic knowledge graph-based achievement quoted quantity prediction method, medium and equipment | |
Kobets et al. | Using an evolutionary algorithm to improve investment strategies for industries in an economic system | |
CN110298506A (en) | A kind of urban construction horizontal forecast system | |
Chusyairi et al. | Optimization of exponential smoothing method using genetic algorithm to predict e-report service | |
Tembine | Mean field stochastic games: Convergence, Q/H-learning and optimality | |
CN111695739B (en) | Load prediction method, system and equipment | |
Leathart et al. | Temporal probability calibration | |
CN111666991A (en) | Convolutional neural network-based pattern recognition method and device and computer equipment | |
CN115622034A (en) | Internet of things meter power load identification method based on improved genetic neural network | |
CN113822441B (en) | Decision model training method, device, terminal equipment and storage medium | |
CN109902870A (en) | Electric grid investment prediction technique based on AdaBoost regression tree model | |
US20210342691A1 (en) | System and method for neural time series preprocessing | |
CN108471362B (en) | Resource allocation prediction technique and device | |
CN113987261A (en) | Video recommendation method and system based on dynamic trust perception |
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 |