CN115271274B - Short-term daily load prediction method for power system and related equipment - Google Patents
Short-term daily load prediction method for power system and related equipment Download PDFInfo
- Publication number
- CN115271274B CN115271274B CN202211204902.8A CN202211204902A CN115271274B CN 115271274 B CN115271274 B CN 115271274B CN 202211204902 A CN202211204902 A CN 202211204902A CN 115271274 B CN115271274 B CN 115271274B
- Authority
- CN
- China
- Prior art keywords
- daily load
- dimensional
- load array
- prediction
- array
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 99
- 239000013598 vector Substances 0.000 claims abstract description 117
- 239000000203 mixture Substances 0.000 claims abstract description 58
- 230000006870 function Effects 0.000 claims description 41
- 238000012549 training Methods 0.000 claims description 39
- 238000004422 calculation algorithm Methods 0.000 claims description 28
- 230000009467 reduction Effects 0.000 claims description 21
- 238000013507 mapping Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 abstract description 6
- 238000009826 distribution Methods 0.000 description 15
- 238000004364 calculation method Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 11
- 239000011159 matrix material Substances 0.000 description 10
- 150000001875 compounds Chemical class 0.000 description 8
- 230000008901 benefit Effects 0.000 description 7
- 238000004590 computer program Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000003491 array Methods 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 230000004931 aggregating effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- 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
-
- 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)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Power Engineering (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method and related equipment for predicting short-term daily load of a power system, relates to the field of power load prediction, and mainly solves the problem that a method capable of predicting the short-term daily load of the power system according to category is absent at present. The method comprises the following steps: clustering the 2-dimensional daily load array based on a Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined based on a 24-dimensional daily load array; determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, wherein the correlation vector machine model is a predictive model; and determining a daily load prediction result of the target prediction day based on the prediction model. The method is used for the short-term daily load prediction process of the power system.
Description
Technical Field
The invention relates to the field of power load prediction, in particular to a short-term daily load prediction method and related equipment for a power system.
Background
Accurate short-term power load prediction is the basis for a power system to make a day-ahead scheduling plan and is also an important basis for a power generation main body to make a reasonable bidding strategy in a power market environment so as to obtain better power generation benefits. However, the existing technical solutions mainly include two disadvantages: firstly, the existing power short-term load forecasting method usually forecasts according to the time sequence of the load, and is difficult to track the sudden change of the load under extreme conditions; secondly, most of the existing improved technical schemes focus on improving the load prediction precision, the key index of the calculation efficiency is not considered, the calculation efficiency of the machine learning algorithm generally decreases exponentially along with the improvement of the precision requirement, and the time requirement of the online dynamic load prediction of the current power system is difficult to meet.
Disclosure of Invention
In view of the above problems, the present invention provides a method for predicting a short-term daily load of an electrical power system and related equipment, and mainly aims to solve the problem that a method for predicting a short-term daily load of an electrical power system according to a "class" is absent at present.
In order to solve at least one technical problem, in a first aspect, the present invention provides a method for predicting a short-term daily load of an electrical power system, the method including:
clustering the 2-dimensional daily load array based on a Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined based on a 24-dimensional daily load array;
determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, wherein the correlation vector machine model is a prediction model;
and determining a daily load prediction result of the target prediction day based on the prediction model.
Optionally, the method further includes:
acquiring the 24-dimensional daily load data in a statistical period to determine a 24-dimensional daily load array, wherein the 24-dimensional daily load data is used for representing 24-hour daily load data of one day;
and reducing the 24-dimensional daily load array into the 2-dimensional daily load array based on a multidimensional scale method, wherein the 24-dimensional daily load array and the 2-dimensional daily load array have a mapping relation.
Optionally, the clustering the 2-dimensional daily load array based on the gaussian mixture model to obtain the target 2-dimensional daily load array includes:
determining the Gaussian mixture model based on the 2-dimensional daily load array and the probability density function;
determining the quantity of the components of the Gaussian mixture model based on a Bayesian criterion, wherein the quantity of the components of the Gaussian mixture model is equal to the quantity of the clustering centers;
determining a log-likelihood estimation function based on the Gaussian mixture model;
and iteratively solving the log likelihood estimation function through an EM algorithm and the number of the clustering centers to obtain a target 2-dimensional daily load array.
Optionally, the determining a relevance vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array includes:
based on the mapping relation between the 24-dimensional daily load array and the 2-dimensional daily load array, acquiring a training feature vector of each cluster of data in the 2-dimensional daily load array from the 24-dimensional daily load array, wherein the training feature vector is meteorological data and load data in a period of time before a training day;
and establishing a relevant vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array.
Optionally, the establishing a relevant vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array includes:
and establishing the correlation vector machine model based on the training characteristic vector, the function weight and the Gaussian white noise.
Optionally, the determining a daily load prediction result of the target prediction day based on the prediction model includes:
determining a target correlation vector machine model corresponding to a cluster in the 2-dimensional daily load array to which the target prediction day belongs;
and solving the target correlation vector machine model based on an EM algorithm and a prediction characteristic vector to determine a daily load prediction result of a target prediction day, wherein the prediction characteristic vector is meteorological data and load data in a period of time before the target prediction day.
Optionally, the method further includes:
determining a prediction error based on the daily load prediction result and the daily load actual result;
and analyzing and evaluating the prediction error based on the average absolute percentage error and the root-mean-square error.
In a second aspect, an embodiment of the present invention further provides a device for predicting short-term daily load of an electrical power system, including:
the clustering unit is used for clustering the 2-dimensional daily load array based on a Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined based on a 24-dimensional daily load array;
a determining unit, configured to determine a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, where the correlation vector machine model is a prediction model;
and a second determination unit configured to determine a daily load prediction result of the target prediction day based on the prediction model.
In order to achieve the above object, according to a third aspect of the present invention, there is provided a computer-readable storage medium including a stored program, wherein the steps of the power system short-term daily load prediction method described above are implemented when the program is executed by a processor.
In order to achieve the above object, according to a fourth aspect of the present invention, there is provided an electronic device comprising at least one processor, and at least one memory connected to the processor; the processor is used for calling the program instructions in the memory and executing the steps of the power system short-term daily load prediction method.
By means of the technical scheme, for the problem that a method capable of predicting the short-term daily load of the power system according to the category is absent at present, the method and the related equipment cluster the 2-dimensional daily load array based on the Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined based on the 24-dimensional daily load array; determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, wherein the correlation vector machine model is a prediction model; and determining a daily load prediction result of the target prediction day based on the prediction model. In the scheme, the acquired data bulkiness and multidimensional degree of the 24-dimensional daily load array can increase the cost of subsequent clustering operation and reduce the clustering efficiency, and meanwhile, a high-dimensional clustering result is difficult to analyze, observe and interpret, so that the method determines the 2-dimensional daily load array based on the 24-dimensional daily load array, improves the efficiency of subsequent calculation while ensuring that the data accuracy is not greatly lost, then clusters the 2-dimensional daily load array through a Gaussian mixture model, thereby determining the clusters corresponding to each 2-dimensional daily load array and the 24-dimensional daily load array, establishing a corresponding relevant vector machine model for each cluster, and facilitating the subsequent prediction based on the cluster to which a target prediction day belongs.
Accordingly, the device, the equipment and the computer-readable storage medium for predicting the short-term daily load of the power system provided by the embodiment of the invention also have the technical effects.
The above description is only an overview of the technical solutions of the present invention, and the present invention can be implemented in accordance with the content of the description so as to make the technical means of the present invention more clearly understood, and the above and other objects, features, and advantages of the present invention will be more clearly understood.
Drawings
Various additional advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
fig. 1 is a schematic flow chart illustrating a method for predicting a short-term daily load of an electric power system according to an embodiment of the present invention;
fig. 2 shows a 2-dimensional data point coordinate diagram after dimension reduction in a short-term daily load prediction method for an electric power system according to an embodiment of the present invention;
fig. 3 shows a 2-dimensional data point coordinate graph after clustering in the short-term daily load prediction method for the power system provided by the embodiment of the invention;
fig. 4 shows a graph of the prediction results of the four-season daily load of the area according to the time sequence and the category when the short-term daily load prediction method of the power system is provided by the embodiment of the invention;
FIG. 5 is a graph showing a comparison between the four-season class prediction and the chronological prediction error MAPE in a short-term daily load prediction method for an electric power system according to an embodiment of the present invention;
FIG. 6 is a graph showing comparison between the prediction of four seasons by classes and the prediction error RMSE in time sequence in the region when the short-term daily load prediction method of the power system is provided by the embodiment of the invention;
FIG. 7 is a graph showing comparison between the algorithm class prediction and the time calculation in time sequence for each class in the region when the short-term daily load prediction method for the power system is provided in the embodiment of the invention;
fig. 8 is a block diagram schematically illustrating a short-term daily load prediction apparatus for an electrical power system according to an embodiment of the present invention;
fig. 9 is a block diagram schematically illustrating a composition of an electric power system short-term daily load prediction electronic device according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
In order to solve the problem that a method for predicting the short-term daily load of the power system according to the class is absent at present, an embodiment of the present invention provides a method for predicting the short-term daily load of the power system, as shown in fig. 1, the method includes:
s101, clustering a 2-dimensional daily load array based on a Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined based on a 24-dimensional daily load array;
illustratively, a Gaussian Mixture Model (GMM) completes clustering by selecting a component maximization posterior probability, the posterior probability of each data point represents the possibility of belonging to various classes, but does not judge that it completely belongs to a certain class, because the Gaussian Mixture Model is a probability Model with the fastest learning speed, the principle is that an optimal mixed multidimensional Gaussian distribution Model is constructed by fitting an input data set, most data set distributions are actually Gaussian distributions, even if the original data set distributions are not Gaussian distributions, the distributions tend to be Gaussian distributions along with the increase of sample size according to the central limit theorem, so that the larger the data set size is, the better the fitting effect of the Gaussian Mixture Model is, and because the scheme adopts load data of each day in a statistical cycle for modeling, the data volume is huge, the Gaussian Mixture Model algorithm is more suitable for the method than other clustering algorithms. The method comprises the steps of firstly clustering 2-dimensional daily load arrays in a statistical period based on a Gaussian mixture model, thereby aggregating 2-dimensional daily load distances with high data feature point contact degrees, and determining which cluster in a target 2-dimensional daily load array each daily load data in the 2-dimensional daily load arrays belongs to.
Exemplarily, because the acquired data of the 24-dimensional daily load array is huge and multidimensional, the cost of subsequent clustering operation can be increased, the clustering efficiency is reduced, and meanwhile, the high-dimensional clustering result is difficult to analyze, observe and interpret, the method determines the 2-dimensional daily load array based on the 24-dimensional daily load array, so that the efficiency of subsequent calculation is improved while the data precision is ensured not to be greatly lost.
S102, determining a relevant vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, wherein the relevant vector machine model is a prediction model;
illustratively, a Relevance Vector Machine (RVM) is a bayesian sparse kernel method for regression and classification problems, and since it avoids that a kernel function of a Support Vector Machine (SVM) is a positive definite kernel, the Relevance Vector Machine generally generates a more sparse model compared with the Support Vector Machine, so that the speed on a test set is faster, and since the method determines a prediction model of the Relevance Vector Machine based on a target 2-dimensional daily load array, the method improves the prediction speed while ensuring the prediction accuracy because the Relevance Vector Machine has the advantages of sparsity, probability and freely selectable kernel functions.
And S103, determining a daily load prediction result of the target prediction day based on the prediction model.
In an exemplary manner, a prediction model is constructed by using a correlation vector machine with high prediction speed, high generalization capability and high prediction precision, and a daily load prediction result of a target prediction day is determined based on the prediction model, so that the demand of on-line real-time dynamic prediction of short-term daily load of a power system can be met while the prediction precision is improved.
By means of the technical scheme, the method for predicting the short-term daily load of the power system, provided by the invention, has the advantages that for the problem that a method capable of predicting the short-term daily load of the power system according to the category is lacked at present, the method is used for clustering a 2-dimensional daily load array based on a Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined based on a 24-dimensional daily load array; determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, wherein the correlation vector machine model is a prediction model; and determining a daily load prediction result of the target prediction day based on the prediction model. In the scheme, the acquired data bulkiness and multidimensional degree of the 24-dimensional daily load array can increase the cost of subsequent clustering operation and reduce the clustering efficiency, and meanwhile, a high-dimensional clustering result is difficult to analyze, observe and interpret, so that the method determines the 2-dimensional daily load array based on the 24-dimensional daily load array, improves the efficiency of subsequent calculation while ensuring that the data accuracy is not greatly lost, then clusters the 2-dimensional daily load array through a Gaussian mixture model, thereby determining the clusters corresponding to each 2-dimensional daily load array and the 24-dimensional daily load array, establishing a corresponding relevant vector machine model for each cluster, and facilitating the subsequent prediction based on the cluster to which a target prediction day belongs.
In one embodiment, the method further comprises:
acquiring the 24-dimensional daily load data in a statistical period to determine a 24-dimensional daily load array, wherein the 24-dimensional daily load data are used for representing 24-hour daily load data of one day;
and reducing the 24-dimensional daily load array into the 2-dimensional daily load array based on a multidimensional scale method, wherein the 24-dimensional daily load array and the 2-dimensional daily load array have a mapping relation.
Illustratively, the method obtains the electricity load data of each day in the statistical period, namely the 24-dimensional daily load array, because the obtained data is huge and multidimensional, the cost of subsequent clustering operation is increased, the clustering efficiency is reduced, and meanwhile, the high-dimensional clustering result is difficult to analyze, observe and explain, so the method introduces a dimension reduction technology. Therefore, before using the clustering algorithm, the 24-dimensional daily load array is reduced to the 2-dimensional daily load array based on a Multidimensional Scaling (MDS), which is a classical data dimension reduction method, and when we can only obtain a similarity matrix between samples, the relative positions of the samples in a low-dimensional space can be reconstructed based on only the known distances between the samples in the high-dimensional space, so as to complete dimension reduction.
For example, different data sets have different characteristics, different dimension reduction algorithms need to be applied, and a clustering algorithm and a dimension reduction algorithm may have incompatibility, so that the optimal dimension reduction technology applied to the clustering algorithm cannot necessarily obtain optimal output, different judgment criteria cannot be applied to all the data sets, and the dimension reduction algorithm needs to be determined according to actual needs. The method is characterized in that dimension reduction is respectively carried out on the 24-dimensional daily load array through an experiment based on three dimension reduction methods of t-SNE, PCA and MDS, and finally the determined dimension reduction result can reflect that the dimension reduction rate of the MDS (multi-dimensional scale method) is medium, but compared with the t-SNE and PCA dimension reduction technologies, the dimension reduction loss and the dimension reduction quality of the MDS are better, so that the method selects the multi-dimensional scale method to reduce the dimension of the 24-dimensional daily load array, and the determined 2-dimensional daily load array after dimension reduction is data which can represent the 24-dimensional daily load array most, the data size is reduced while the data characteristics and quality are ensured, the cost is saved for subsequent clustering, and the efficiency is improved.
Illustratively, the method applies to a 24-dimensional raw normalized daily load matrix, i.e., a 24-dimensional daily load array
N is the daily load days in the statistical period, and Euclidean distance is selected as the similarity measurement index among 24-dimensional daily load data, so that the distance matrix of N24-dimensional samples in 24-dimensional space
Wherein the first
Line of
Elements of a column
Is a sample
To
The Euclidean distance calculation method comprises the following steps:
in the above-mentioned formula, the compound of formula,
and
are respectively
And
to (1) a
And (5) dimension elements.
Assume the 2-dimensional data after dimension reduction is
Then the distance matrix of N2-dimensional samples in 2-dimensional space
Wherein a first
Line of
Elements of a column
Is a sample
To
To construct an intermediate computational matrix that is easy to solve
Let us order
，
Wherein a first
Line for mobile communication terminal
The elements of the column are represented as
The calculation method comprises the following steps:
to pair
Performing characteristic decomposition:
in the above-mentioned formula, the compound has the following formula,
diagonal matrix formed for eigenvalues, eigenvalues
，
Is a feature vector matrix.
Then construct
，
Is its corresponding eigenvector matrix.
Can obtain
。
In an embodiment, the clustering the 2-dimensional daily load array based on the gaussian mixture model to obtain the target 2-dimensional daily load array includes:
determining the Gaussian mixture model based on the 2-dimensional daily load array and the probability density function;
determining the quantity of the components of the Gaussian mixture model based on a Bayesian criterion, wherein the quantity of the components of the Gaussian mixture model is equal to the quantity of the clustering centers;
determining a log-likelihood estimation function based on the Gaussian mixture model;
and iteratively solving the log likelihood estimation function through an EM algorithm and the number of the clustering centers to obtain a target 2-dimensional daily load array.
Illustratively, the probability density function of the gaussian mixture model described above needs to be expressed first:
in the above-mentioned formula, the compound of formula,
to be the total number of cluster centers,
is a specific cluster center of the cluster centers,
to represent the first
Weights of the individual components;
is the mean of the gaussian distribution;
the probability density function of the Gaussian mixture model is mainly composed of parameter sets for the covariance of Gaussian distribution
And (4) forming.
Illustratively, in the case of known data samples, i.e. the 2-dimensional daily load array and the expression of the gaussian mixture model, the number of cluster centers can be determined by using a Bayesian criterion, which is one of the commonly used methods for determining the optimal number of clusters of the gaussian mixture model (BIC), i.e. the 2-dimensional daily load array and the expression of the gaussian mixture model
The number of clustering clusters is the optimal number of clustering centers of the method. After the number of the clustering centers is determined, the 2-dimensional daily load arrays can be clustered to obtain a target 2-dimensional daily load array, so that 2-dimensional daily load distances with high data feature point contact ratio are aggregated.
For example, in clustering, a gaussian mixture model faced with unknown parameters requires the estimation of gaussian mixture model parameters since it is not possible to determine from which potential component each of the sites originated. The probability that the gaussian mixture model generates a data point (i.e., the likelihood estimation function) needs to be solved first but the logarithm of the likelihood estimation function, i.e., the log likelihood estimation function, is often taken due to the computational complexity. The log-likelihood estimation function determined based on the gaussian mixture model is:
exemplarily, an EM algorithm (expectation-maximization algorithm) is a common method for estimating a gaussian mixture model, and the basic idea of the EM algorithm is to solve a maximum likelihood estimation of a model distribution parameter by introducing hidden variables, and then repeatedly iterate an hidden variable expectation formula and a model distribution parameter reestimation formula until likelihood function values converge.
Illustratively, the Gaussian mixture model is mainly composed of parameter sets
The data sample, namely the implicit variable of each sample in the 2-dimensional daily load array is
，
Is an optimized approximate quantity obtained after the overall data observation, and the value can be expressed as
Specifically, the method can be determined after the optimization solution of the EM algorithm. And solving the log-likelihood estimation function by adopting an EM algorithm, wherein iteration is mainly divided into an expectation Step (E-Step) and a maximization Step (M-Step).
E-step: calculating the expected value of the sample hidden variable according to the current parameter value, wherein the formal expression is as follows:
m-step: according to the hidden variables of the current sample, solving the maximum likelihood estimation of the parameters can be expressed as:
and based on the introduced hidden variables, the M-Step solves the parameters of the updated model according to the result value of the E-Step. And repeating the E-Step and the M-Step until the parameters or the log-likelihood function are converged to obtain the optimal GMM parameters, namely finishing the classification of the dimension-reducing daily load data, and aggregating the 2-dimensional daily load data with higher data feature point coincidence degree to form a target 2-dimensional daily load array.
In one embodiment, the determining a relevance vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array includes:
based on the mapping relation between the 24-dimensional daily load array and the 2-dimensional daily load array, acquiring a training feature vector of each cluster of data in the 2-dimensional daily load array from the 24-dimensional daily load array, wherein the training feature vector is meteorological data and load data in a period of time before a training day;
and establishing a relevant vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array.
Exemplarily, a target 2-dimensional daily load array is determined after clustering is completed, that is, the target 2-dimensional daily load array is correspondingly divided into a plurality of clusters, each cluster includes a plurality of 2-dimensional daily load data, and each 2-dimensional daily load data corresponds to one 24-dimensional daily load data, so that the 24-dimensional daily load array and the 2-dimensional daily load array have a mapping relationship, and the 24-dimensional daily load array and the target 2-dimensional daily load array also have a mapping relationship. It should be noted that the dimension reduction of the 24-dimensional daily load array to form the 2-dimensional daily load array is performed by the method to facilitate subsequent clustering, the clustering of the 2-dimensional daily load array to form the target 2-dimensional daily load array is performed to determine which cluster of the related vector machine model needs to be established, and data used in specific establishment of the related vector machine model is derived from the 24-dimensional daily load array to ensure the accuracy of the established model.
Illustratively, first, the 24-dimensional daily load corresponding to each cluster in the 2-dimensional daily load array is requiredObtaining training feature vectors for input training in an array
The load data of (a) includes: load data, daily maximum, minimum, average load data, and meteorological data such as daily maximum, minimum, average temperature, average humidity, daily air pressure, and daily rainfall are trained for a period of time prior to the day (e.g., the previous 7 days), and a correlation vector machine model is then built based thereon.
In an embodiment, the establishing a correlation vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array includes:
and establishing the correlation vector machine model based on the training characteristic vector, the function weight and the Gaussian white noise.
Illustratively, from input feature vectors
Weight of the basis function
And white gaussian noise
Expresses the initial correlation vector machine model as follows:
in the above-mentioned formula, the compound of formula,
the predicted value of the representative daily load, above
Representing white Gaussian noise with an average value of 0 and a variance of
The above-mentioned
Can be determined by a fixed non-linear basis function given by the kernel
Represents:
the likelihood function is a function of the parameters of the statistical model, the observation matrix of the data inputted in N dimensions
The calculated likelihood function of (a) is:
function weight of the likelihood function
The prior probability distribution of (a) is:
in the above-mentioned formula, the compound has the following formula,
representing corresponding parameters
The accuracy of (2).
Function weight of the likelihood function
The posterior probability distribution of (a) is:
in the above formula, the function weight of the likelihood function
Mean value of
Sum variance
Comprises the following steps:
in the above-mentioned formula, the compound of formula,
is composed of
The matrix is designed in a dimension mode,
，
。
therefore, the construction of the relevant vector machine model based on the training feature vector, the function weight and the Gaussian white noise is completed.
In an embodiment, the determining the daily load prediction result of the target prediction day based on the prediction model includes:
determining a target relevance vector machine model corresponding to a cluster in the 2-dimensional daily load array to which the target prediction day belongs;
and solving the target related vector machine model based on an EM algorithm and a prediction characteristic vector to determine a daily load prediction result of a target prediction day, wherein the prediction characteristic vector is meteorological data and load data in a period of time before the target prediction day.
For example, the learning process of the correlation vector machine model can be converted into parameters based on the above formula
And
by estimating the parameters using EM algorithm
And
optimum value of (2)
And
the calculation formula is as follows:
for example, after determining the correlation vector machine model corresponding to each cluster in the target 2-dimensional daily load array, a target prediction day to be predicted is obtained, a cluster to which the target prediction day belongs, that is, the target correlation vector machine model required for prediction of the target prediction day is determined, and the prediction feature vector is obtained based on the 24-dimensional daily load array again
The load data of (a) includes: target forecast daily period (e.g. 7 days) load data, daily maximum, minimumAverage load data, and weather data such as daily maximum, minimum, average air temperature, average humidity, daily air pressure, and daily rainfall. For new input feature vectors, one can calculate
The predicted distribution of (c) is:
in the above-mentioned formula, the compound of formula,
function weights as likelihood functions
Mean value of
Is about to be transferred
All elements are mirror-inverted around a 45 degree right-lower ray from the row 1, column 1 element to obtain
The transposing of (1).
The variance through the prediction distribution is:
thereby obtaining input feature vectors
Prediction value of corresponding correlation vector machine model
Comprises the following steps:
therefore, a combined prediction model is built by utilizing a related vector mechanism with higher prediction speed, better generalization capability and higher prediction precision, and the process of predicting the short-term daily load of the power system is completed. It is noted that in the prediction using the present method, the load data and meteorological data for a period of time prior to the target prediction day need to be known.
In one embodiment, the method further comprises:
determining a prediction error based on the daily load prediction result and the daily load actual result;
the above prediction error is analyzed and evaluated based on the mean absolute percentage error and the root mean square error.
Illustratively, the method selects the Mean Absolute Percent Error (MAPE) and the Root Mean Square Error (RMSE) for analysis of the prediction error:
in the above-mentioned formula, the compound of formula,
in order to predict the number of samples,
in order to predict the value of the load,
is the actual load value. Therefore, the analysis and evaluation of the prediction result are completed, the accuracy of the prediction result can be conveniently observed by workers, and the model is further improved on the basis of the accuracy.
For example, an example analysis is performed using 8760 hours of load data for a year in a certain area:
the original 24-dimensional daily load data of the region is reduced to 2-dimensional data through a multi-dimensional scale method, a coordinate diagram of the 2-dimensional data points after the dimension reduction is shown in fig. 2, and horizontal and vertical coordinates in fig. 2 are power loads. The optimal clustering number of the Gaussian mixture model is selected to be 7 types according to the Bayesian criterion, the result of the Gaussian mixture model clustering of the 2-dimensional data points and the corresponding load curve clustering result are shown in FIG. 3, the horizontal and vertical coordinates in FIG. 3 are power loads, and the types 1-7 are that the 2-dimensional data are clustered into 7 clusters. In order to verify the superiority of prediction according to the 'day type' compared with prediction according to the traditional time, 6 advanced prediction algorithms are adopted to randomly select 10 days in spring, summer, autumn and winter in the area, and comparative analysis is carried out, specifically, 5 months and 1 day to 5 months and 10 days are selected in spring, 8 months and 22 days to 8 months and 31 days are selected in summer, 11 months and 9 days to 11 months and 18 days are selected in autumn, and 2 months and 19 days to 2 months and 28 days are selected in winter. Taking summer as an example for explanation, selecting 8 months 22-8 months 31 days as a test set, and predicting according to the time sequence, namely predicting by using the rest days of 6-8 months as a training set; and (4) predicting according to classes, namely judging which of 7 classes classified by the clustering data set belongs to from 8 months 22 days to 8 months 31 days by using a classifier, putting the classes into a corresponding training set for prediction, and establishing the classifier by using a correlation vector machine model. The prediction results of the four-season daily load in the province according to time sequence and category are shown in the following figures 4, 5 and 6, as can be seen from the figure 4, the method provided by the scheme can better track the sudden change of the load curve at the time of the upper steep peak and the lower steep peak of the daily load change according to the prediction of the RVM according to the time sequence, the abscissa of each graph in the figure 4 is time, and the ordinate is power load. And as can be seen from fig. 5 and 6, the prediction accuracy of the prediction by category in four seasons of the region is improved compared with the prediction by time sequence, wherein the prediction by category has the highest prediction accuracy, the average MAPE is 3.38%, the average RMSE is 44.54MW, and fig. 7 lists the calculation time of various algorithms adopting prediction by category and prediction by time sequence, thereby verifying the superiority and effectiveness of the method of the scheme.
Further, as an implementation of the method shown in fig. 1, an embodiment of the present invention further provides a device for predicting a short-term daily load of an electric power system, which is used to implement the method shown in fig. 1. The embodiment of the apparatus corresponds to the embodiment of the method, and for convenience of reading, details in the embodiment of the apparatus are not repeated one by one, but it should be clear that the apparatus in the embodiment can correspondingly implement all the contents in the embodiment of the method. As shown in fig. 8, the apparatus includes: a clustering unit 21, a determining unit 22 and a second determining unit 23, wherein
A clustering unit 21, configured to cluster the 2-dimensional daily load array based on a gaussian mixture model to obtain a target 2-dimensional daily load array, where the 2-dimensional daily load array is determined based on a 24-dimensional daily load array;
a determining unit 22, configured to determine a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, where the correlation vector machine model is a prediction model;
a second determination unit 23, configured to determine a daily load prediction result of the target prediction day based on the prediction model.
Exemplarily, the above unit is further configured to:
acquiring the 24-dimensional daily load data in a statistical period to determine a 24-dimensional daily load array, wherein the 24-dimensional daily load data is used for representing 24-hour daily load data of one day;
and reducing the 24-dimensional daily load array into the 2-dimensional daily load array based on a multidimensional scale method, wherein the 24-dimensional daily load array and the 2-dimensional daily load array have a mapping relation.
For example, the clustering the 2-dimensional daily load array based on the gaussian mixture model to obtain the target 2-dimensional daily load array includes:
determining the Gaussian mixture model based on the 2-dimensional daily load array and the probability density function;
determining the quantity of the components of the Gaussian mixture model based on a Bayesian criterion, wherein the quantity of the components of the Gaussian mixture model is equal to the quantity of the clustering centers;
determining a log-likelihood estimation function based on the Gaussian mixture model;
and iteratively solving the log likelihood estimation function through an EM algorithm and the number of the clustering centers to obtain a target 2-dimensional daily load array.
For example, the determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array includes:
acquiring a training feature vector of each cluster of data in the 2-dimensional daily load array from the 24-dimensional daily load array based on a mapping relation between the 24-dimensional daily load array and the 2-dimensional daily load array, wherein the training feature vector is meteorological data and load data in a period of time before a training day;
and establishing a relevant vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array.
For example, the establishing of the correlation vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array includes:
and establishing the correlation vector machine model based on the training characteristic vector, the function weight and the Gaussian white noise.
For example, the determining the daily load prediction result of the target prediction day based on the prediction model includes:
determining a target correlation vector machine model corresponding to a cluster in the 2-dimensional daily load array to which the target prediction day belongs;
and solving the target related vector machine model based on an EM algorithm and a prediction characteristic vector to determine a daily load prediction result of a target prediction day, wherein the prediction characteristic vector is meteorological data and load data in a period of time before the target prediction day.
Exemplarily, the above unit is further configured to:
determining a prediction error based on the daily load prediction result and the daily load actual result;
and analyzing and evaluating the prediction error based on the average absolute percentage error and the root-mean-square error.
By means of the technical scheme, the power system short-term daily load forecasting device provided by the invention has the advantages that for the problem that a method capable of forecasting the power system short-term daily load according to the category is lacked at present, the 2-dimensional daily load array is clustered on the basis of a Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined on the basis of a 24-dimensional daily load array; determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, wherein the correlation vector machine model is a prediction model; and determining a daily load prediction result of the target prediction day based on the prediction model. In the scheme, because the acquired 24-dimensional daily load array has large data and multiple dimensions, the cost of subsequent clustering operation can be increased, the clustering efficiency is reduced, and meanwhile, a high-dimension clustering result is difficult to analyze, observe and interpret, the method determines the 2-dimensional daily load array based on the 24-dimensional daily load array, improves the efficiency of subsequent calculation while ensuring that the data precision is not greatly lost, then clusters the 2-dimensional daily load array through a Gaussian mixture model, thereby determining the clusters corresponding to each 2-dimensional daily load array and the 24-dimensional daily load array, establishing a corresponding relevant vector machine model for each cluster, facilitating the subsequent prediction based on the cluster to which a target prediction day belongs, improving the traditional load prediction mode trained in time sequence into prediction according to a 'day type', effectively improving the short-term load prediction precision and the load mutation tracking capacity under extreme conditions, and utilizing a relevant vector machine prediction model with higher prediction speed, better optimization capacity and higher prediction precision, and meeting the speed demand of short-term daily load prediction of a power system while improving the prediction.
The processor comprises a kernel, and the kernel calls the corresponding program unit from the memory. One or more than one kernel can be set, the method for predicting the short-term daily load of the power system is realized by adjusting kernel parameters, and the problem that a method for predicting the short-term daily load of the power system according to the category is lacked at present can be solved.
An embodiment of the present invention provides a computer-readable storage medium, where the computer-readable storage medium includes a stored program, and the program, when executed by a processor, implements the power system short-term daily load prediction method.
The embodiment of the invention provides a processor, which is used for running a program, wherein the program executes the method for predicting the short-term daily load of the power system when running.
The embodiment of the invention provides electronic equipment, which comprises at least one processor and at least one memory connected with the processor; the processor is used for calling the program instructions in the memory and executing the power system short-term daily load prediction method
An embodiment of the present invention provides an electronic device 30, as shown in fig. 9, the electronic device includes at least one processor 301, at least one memory 302 connected to the processor, and a bus 303; the processor 301 and the memory 302 complete communication with each other through the bus 303; the processor 301 is configured to call program instructions in the memory to perform the power system short-term daily load prediction method described above.
The intelligent electronic device herein may be a PC, PAD, mobile phone, etc.
The present application further provides a computer program product adapted to perform a program for initializing the following method steps when executed on a flow management electronic device:
clustering the 2-dimensional daily load array based on a Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined based on a 24-dimensional daily load array;
determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, wherein the correlation vector machine model is a prediction model;
and determining a daily load prediction result of the target prediction day based on the prediction model.
Further, the method further comprises:
acquiring the 24-dimensional daily load data in a statistical period to determine a 24-dimensional daily load array, wherein the 24-dimensional daily load data are used for representing 24-hour daily load data of one day;
and reducing the 24-dimensional daily load array into the 2-dimensional daily load array based on a multidimensional scale method, wherein the 24-dimensional daily load array and the 2-dimensional daily load array have a mapping relation.
Further, the clustering the 2-dimensional daily load array based on the gaussian mixture model to obtain the target 2-dimensional daily load array includes:
determining the Gaussian mixture model based on the 2-dimensional daily load array and the probability density function;
determining the quantity of the components of the Gaussian mixture model based on a Bayesian criterion, wherein the quantity of the components of the Gaussian mixture model is equal to the quantity of the clustering centers;
determining a log-likelihood estimation function based on the Gaussian mixture model;
and iteratively solving the log likelihood estimation function through an EM algorithm and the number of the clustering centers to obtain a target 2-dimensional daily load array.
Further, the determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array includes:
acquiring a training feature vector of each cluster of data in the 2-dimensional daily load array from the 24-dimensional daily load array based on a mapping relation between the 24-dimensional daily load array and the 2-dimensional daily load array, wherein the training feature vector is meteorological data and load data in a period of time before a training day;
and establishing a relevant vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array.
Further, the establishing of the relevant vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array includes:
and establishing the correlation vector machine model based on the training characteristic vector, the function weight and the Gaussian white noise.
Further, the determining a daily load prediction result of the target prediction day based on the prediction model includes:
determining a target correlation vector machine model corresponding to a cluster in the 2-dimensional daily load array to which the target prediction day belongs;
and solving the target related vector machine model based on an EM algorithm and a prediction characteristic vector to determine a daily load prediction result of a target prediction day, wherein the prediction characteristic vector is meteorological data and load data in a period of time before the target prediction day.
Further, the method further comprises:
determining a prediction error based on the daily load prediction result and the daily load actual result;
the above prediction error is analyzed and evaluated based on the mean absolute percentage error and the root mean square error.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, electronic devices (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable flow management electronic device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable flow management electronic device, create means for implementing the functions specified in the flow diagram flow or flows and/or block diagram block or blocks.
In a typical configuration, an electronic device includes one or more processors (CPUs), memory, and a bus. The electronic device may also include input/output interfaces, network interfaces, and the like.
The memory may include volatile memory in a computer readable medium, random Access Memory (RAM) and/or nonvolatile memory such as Read Only Memory (ROM) or flash memory (flash RAM), and the memory includes at least one memory chip. The memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer-readable storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage electronic devices, or any other non-transmission medium, that can be used to store information that can be accessed by computing electronic devices. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or electronic device that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or electronic device. Without further limitation, an element defined by the phrases "comprising one of 8230; \8230;" 8230; "does not exclude the presence of additional like elements in a process, method, article, or electronic device that comprises the element.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art to which the present application pertains. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (8)
1. A method for predicting short-term daily load of a power system is characterized by comprising the following steps:
clustering a 2-dimensional daily load array based on a Gaussian mixture model to obtain a target 2-dimensional daily load array, wherein the 2-dimensional daily load array is determined based on a 24-dimensional daily load array, the 24-dimensional daily load array is used for representing 24-hour power load data of at least one day, and the 2-dimensional daily load array is used for representing 2-dimensional characteristic values of the 24-hour power load data of at least one day after dimensionality reduction;
determining a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, wherein the correlation vector machine model is a prediction model;
determining a daily load prediction result of a target prediction day based on the prediction model;
acquiring 24-dimensional daily load data in a statistical period to determine the 24-dimensional daily load array;
reducing the 24-dimensional daily load array into the 2-dimensional daily load array based on a multidimensional scale method, wherein the 24-dimensional daily load array and the 2-dimensional daily load array have a mapping relation;
based on the mapping relation between the 24-dimensional daily load array and the 2-dimensional daily load array, acquiring a training feature vector of each cluster of data in the 2-dimensional daily load array from the 24-dimensional daily load array, wherein the training feature vector is meteorological data and load data in a period of time before a training day;
and establishing a relevant vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array.
2. The method of claim 1, wherein clustering the 2-dimensional daily load array based on the Gaussian mixture model to obtain a target 2-dimensional daily load array comprises:
determining the Gaussian mixture model based on the 2-dimensional daily load array and a probability density function;
determining the number of components of the Gaussian mixture model based on Bayesian criterion, wherein the number of components of the Gaussian mixture model is equal to the number of clustering centers;
determining a log-likelihood estimation function based on the Gaussian mixture model;
and iteratively solving the log likelihood estimation function through an EM algorithm and the number of the clustering centers to obtain a target 2-dimensional daily load array.
3. The method of claim 1, wherein the establishing a correlation vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector for each cluster of data in the 2-dimensional daily load array comprises:
and establishing the correlation vector machine model based on the training feature vector, the function weight and the Gaussian white noise.
4. The method of claim 3, wherein determining a daily load prediction for a target prediction day based on the predictive model comprises:
determining a target relevance vector machine model corresponding to a cluster in the 2-dimensional daily load array to which the target prediction day belongs;
and solving the target correlation vector machine model based on an EM algorithm and a prediction characteristic vector to determine a daily load prediction result of a target prediction day, wherein the prediction characteristic vector is meteorological data and load data in a period of time before the target prediction day.
5. The method of claim 1, further comprising:
determining a prediction error based on the daily load prediction result and the daily load actual result;
and analyzing and evaluating the prediction error based on the average absolute percentage error and the root mean square error.
6. A short-term daily load prediction device for an electric power system,
the system comprises a clustering unit, a storage unit and a processing unit, wherein the clustering unit is used for clustering a 2-dimensional daily load array based on a Gaussian mixture model to obtain a target 2-dimensional daily load array, the 2-dimensional daily load array is determined based on a 24-dimensional daily load array, the 24-dimensional daily load array is used for representing 24-hour power load data of at least one day, and the 2-dimensional daily load array is used for representing 2-dimensional characteristic values of the 24-hour power load data of at least one day after dimensionality reduction;
a determining unit, configured to determine a correlation vector machine model based on the target 2-dimensional daily load array and the 24-dimensional daily load array, where the correlation vector machine model is a prediction model;
a second determination unit configured to determine a daily load prediction result of a target prediction day based on the prediction model;
acquiring 24-dimensional daily load data in a statistical period to determine the 24-dimensional daily load array;
reducing the 24-dimensional daily load array into the 2-dimensional daily load array based on a multidimensional scale method, wherein the 24-dimensional daily load array and the 2-dimensional daily load array have a mapping relation;
based on the mapping relation between the 24-dimensional daily load array and the 2-dimensional daily load array, acquiring a training feature vector of each cluster of data in the 2-dimensional daily load array from the 24-dimensional daily load array, wherein the training feature vector is meteorological data and load data in a period of time before a training day;
and establishing a relevant vector machine model for each cluster of data in the 2-dimensional daily load array based on the training feature vector of each cluster of data in the 2-dimensional daily load array.
7. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored program, wherein the steps of the power system short term daily load prediction method as claimed in any one of claims 1 to 5 are implemented when the program is executed by a processor.
8. An electronic device, comprising at least one processor, and at least one memory coupled to the processor; wherein the processor is configured to invoke program instructions in the memory to perform the steps of the power system short term daily load prediction method as claimed in any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211204902.8A CN115271274B (en) | 2022-09-30 | 2022-09-30 | Short-term daily load prediction method for power system and related equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211204902.8A CN115271274B (en) | 2022-09-30 | 2022-09-30 | Short-term daily load prediction method for power system and related equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115271274A CN115271274A (en) | 2022-11-01 |
| CN115271274B true CN115271274B (en) | 2022-12-27 |
Family
ID=83758126
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211204902.8A Active CN115271274B (en) | 2022-09-30 | 2022-09-30 | Short-term daily load prediction method for power system and related equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115271274B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109409407A (en) * | 2018-09-19 | 2019-03-01 | 西安理工大学 | A kind of industry monitoring data clustering method based on LE algorithm |
| CN109871860A (en) * | 2018-11-02 | 2019-06-11 | 湖南大学 | A kind of daily load curve dimensionality reduction clustering method based on core principle component analysis |
| CN111160625A (en) * | 2019-12-10 | 2020-05-15 | 中铁电气化局集团有限公司 | Power load prediction method, power load prediction device, computer equipment and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11901045B2 (en) * | 2019-01-15 | 2024-02-13 | International Business Machines Corporation | Machine learning framework for finding materials with desired properties |
-
2022
- 2022-09-30 CN CN202211204902.8A patent/CN115271274B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109409407A (en) * | 2018-09-19 | 2019-03-01 | 西安理工大学 | A kind of industry monitoring data clustering method based on LE algorithm |
| CN109871860A (en) * | 2018-11-02 | 2019-06-11 | 湖南大学 | A kind of daily load curve dimensionality reduction clustering method based on core principle component analysis |
| CN111160625A (en) * | 2019-12-10 | 2020-05-15 | 中铁电气化局集团有限公司 | Power load prediction method, power load prediction device, computer equipment and storage medium |
Non-Patent Citations (4)
| Title |
|---|
| 基于核主成分分析和极限学习机的短期电力负荷预测;董浩等;《电子测量与仪器学报》;20180115(第01期);全文 * |
| 基于相空间重构的KPCA-HS-RVM网络流量预测模型;熊文真等;《荆楚理工学院学报》;20161225(第06期);全文 * |
| 实时电价条件下的短期负荷预测;刘梦轩等;《现代电力》;20131010(第05期);全文 * |
| 实时电价条件下的短期负荷预测研究;黄小耘等;《自动化与仪器仪表》;20160525(第05期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115271274A (en) | 2022-11-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111199016B (en) | Daily load curve clustering method for improving K-means based on DTW | |
| CN109871860B (en) | Daily load curve dimension reduction clustering method based on kernel principal component analysis | |
| CN110363354B (en) | Wind power prediction method for wind farm, electronic device and storage medium | |
| CN110674636B (en) | Power consumption behavior analysis method | |
| Pinto et al. | Short-term wind speed forecasting using support vector machines | |
| CN113361785A (en) | Power distribution network short-term load prediction method and device, terminal and storage medium | |
| CN114925767A (en) | Scene generation method and device based on variational self-encoder | |
| CN114676923A (en) | Method and device for predicting generated power, computer equipment and storage medium | |
| CN115905904A (en) | Line loss abnormity evaluation method and device for power distribution network feeder line | |
| CN115759590A (en) | Industrial user demand response potential calculation method based on electricity consumption behavior characteristics | |
| CN113344589B (en) | Intelligent identification method for collusion behavior of power generation enterprise based on VAEGMM model | |
| CN115271274B (en) | Short-term daily load prediction method for power system and related equipment | |
| CN112330051A (en) | Short-term load prediction method based on Kmeans and FR-DBN | |
| CN116911459A (en) | Multi-input multi-output ultra-short-term power load prediction method suitable for virtual power plant | |
| CN116308456A (en) | User electricity consumption behavior analysis method, system, device and medium in big data environment | |
| CN116304295A (en) | User energy consumption portrait analysis method based on multivariate data driving | |
| CN115687948A (en) | Power special transformer user unsupervised classification method based on load curve | |
| CN114723147A (en) | New energy power prediction method based on improved wavelet transform and neural network | |
| Si et al. | Fair hierarchical clustering of substations based on Gini coefficient | |
| CN118133063A (en) | User electricity behavior feature analysis method and system based on demand response | |
| CN113673579B (en) | Small sample-based electricity load classification algorithm | |
| CN118070134B (en) | Method and system for predicting electricity consumption of users in park | |
| CN110932265B (en) | Static voltage stability evaluation method and system | |
| CN115526430A (en) | Load interval prediction method, system and medium for multi-distance clustering and information aggregation | |
| CN115907235A (en) | Photovoltaic power ultra-short term nonparametric probability prediction method based on fuzzy sample particles |
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 |