CN115730740A - Transformer area level power short-term load prediction method and system - Google Patents
Transformer area level power short-term load prediction method and system Download PDFInfo
- Publication number
- CN115730740A CN115730740A CN202211527225.3A CN202211527225A CN115730740A CN 115730740 A CN115730740 A CN 115730740A CN 202211527225 A CN202211527225 A CN 202211527225A CN 115730740 A CN115730740 A CN 115730740A
- Authority
- CN
- China
- Prior art keywords
- load
- low
- voltage distribution
- sequence
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000009826 distribution Methods 0.000 claims abstract description 112
- 239000011159 matrix material Substances 0.000 claims abstract description 41
- 238000010586 diagram Methods 0.000 claims abstract description 39
- 238000012549 training Methods 0.000 claims abstract description 23
- 230000009466 transformation Effects 0.000 claims abstract description 15
- 230000004931 aggregating effect Effects 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims description 23
- 238000013527 convolutional neural network Methods 0.000 claims description 13
- 230000004913 activation Effects 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 7
- 238000000547 structure data Methods 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 230000002776 aggregation Effects 0.000 claims description 4
- 238000004220 aggregation Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 abstract description 5
- 230000006399 behavior Effects 0.000 abstract description 5
- 238000013277 forecasting method Methods 0.000 abstract description 3
- 238000004590 computer program Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- 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
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a transformer area level power short-term load forecasting method and a transformer area level power short-term load forecasting system, and belongs to the technical field of power system load forecasting. The method comprises the following steps: according to a three-phase load sequence of the low-voltage distribution transformer, establishing a characteristic sequence and an adjacent matrix of corresponding nodes, and constructing a space-time diagram data and diagram data sample facing to low-voltage distribution station load prediction; constructing corresponding multi-source condition data characteristics based on the multi-source condition data; training a condition self-adaptive space-time synchronization map convolution neural network (CASSTGCN) model according to the constructed space-time map data and the multi-source condition data characteristics; applying the trained CASSTGCN model to predict a predicted value of the low-voltage distribution transformer three-phase load; and aggregating the load predicted values of the distribution and transformation three phases to obtain the total load predicted value of the low-voltage distribution substation at the next moment. The method effectively solves the problem that the load prediction of the low-voltage distribution station area ignores potential spatial correlation existing between power utilization behaviors of users in the station area at present, and obviously improves the precision of the load prediction of the low-voltage distribution station area.
Description
Technical Field
The invention belongs to the technical field of power system load prediction, and particularly relates to a transformer area level power short-term load prediction method and system.
Background
At present, short-term load forecasting research mainly focuses on provincial or urban regional power grids, and load forecasting with a low-voltage distribution transformer area as a research object faces the problems of small power supply area, large power load fluctuation, insufficient stability of a forecasting result and the like, so that a traditional forecasting method based on coarse-grained data is difficult to be suitable for platform area level load forecasting. With the popularization of distribution transformer online monitoring equipment, the problem of acquisition and storage of mass platform area load data is solved, so that platform area level load prediction based on fine-grained data has a good data basis.
The existing platform region level load prediction method is mainly based on algorithms such as a random forest, a support vector machine, an artificial neural network and a long-short term memory neural network to construct a prediction model. However, the above load prediction methods focus on mining the time correlation of the power load sequence of the distribution area, but ignore the potential spatial correlation existing between the power consumption behaviors of users in the distribution area (for example, in the same distribution area, users share the same geographical space, meteorological conditions, holiday information, power price policy and other comprehensive factors), so that the distribution area level load prediction accuracy is difficult to further improve to some extent.
Disclosure of Invention
The invention mainly aims to overcome the defects and shortcomings in the prior art, provides a station level power short-term load forecasting method and system, effectively solves the problem that potential spatial correlation exists between power utilization behaviors of users in a station is neglected in low-voltage distribution station load forecasting at present, and remarkably improves the accuracy of low-voltage distribution station load forecasting.
In order to achieve the purpose, the invention adopts the following technical scheme:
according to one aspect of the invention, the invention provides a platform level power short-term load prediction method, which comprises the following steps:
according to a three-phase load sequence of the low-voltage distribution transformer, establishing a characteristic sequence and an adjacent matrix of corresponding nodes, and constructing a space-time diagram data and diagram data sample facing to low-voltage distribution station load prediction;
constructing corresponding multi-source condition data characteristics based on multi-source condition data, wherein the multi-source condition data comprise week types, time information, holidays, temperatures, humidity, wind speeds and air pressures;
training a condition self-adaptive space-time synchronization map convolutional neural network (CASSTGCN) model according to the constructed space-time map data and the multi-source condition data characteristics, and excavating a load mode of a low-voltage distribution substation area;
predicting the low-voltage distribution transformer three-phase load by applying the trained CASSTGCN model to respectively obtain the load prediction values of the distribution transformer three phases at the next moment;
and aggregating the load predicted values of the three phases of the distribution transformer to obtain a total load predicted value of the low-voltage distribution substation area at the next moment.
Preferably, the establishing of the characteristic sequence and the adjacency matrix of the corresponding node according to the three-phase load sequence of the low-voltage distribution transformer and the constructing of the spatio-temporal graph data and the graph data sample facing the load prediction of the low-voltage distribution station area include:
taking the three-phase load sequences as the characteristic sequences of each node of the graph structure data, and calculating the correlation coefficient among the phase load sequences;
constructing an adjacency matrix A according to the correlation coefficients, wherein,
The graph data samples are:
wherein,
in order to input the samples, the method comprises the following steps of,
in order to output the samples, the samples are,
a graph signal matrix formed of node features for time t, A ij Is the element adjacent to the ith row and the jth column of the matrix A, xi is the threshold value, rho ij Representing a load sequence X i And X j Pearson's correlation coefficient, COV (X) i ,X j ) Represents X i And X j The covariance of (a) of (b),
and
respectively represent X i And X j Standard deviation of (2).
Preferably, the constructing the corresponding multi-source condition data feature based on the multi-source condition data includes:
respectively constructing a week type sequence W, a time index sequence D, a holiday sign sequence H, a temperature characteristic sequence E, a humidity characteristic sequence M, a wind speed characteristic sequence P and an air pressure characteristic sequence Q, wherein at the time t, a condition data characteristic sequence L consists of the 7 sequences and is represented as follows:
L＝[W,D,H,E,M,P,Q]。
preferably, the training condition adaptive space-time synchronization map convolutional neural network CASSGCN model comprises:
setting parameters of the CASSGCN model, wherein the parameters comprise: the method comprises the steps of historical load sequence input length, condition data characteristic dimension, input transformation layer dimension, space-time embedding layer dimension, graph volume layer number, sliding window length, graph volume layer activation function, output mapping layer dimension, output mapping layer activation function, learning rate, loss function, learning attenuation rate, batch scale and training period.
Preferably, the aggregating the load predicted values of the distribution transformation three phases to obtain a total load predicted value of the low-voltage distribution substation at the next time includes:
the calculation formula of the total load predicted value is as follows:
in the formula, P next The total load prediction value of the low-voltage distribution station area is obtained,
and the load predicted values of the three phases A, B and C of the low-voltage distribution transformer at the next future moment are predicted according to the CASSTGCN model.
According to another aspect of the present invention, there is also provided a station level power short-term load prediction system, the system comprising:
the construction module is used for establishing a characteristic sequence and an adjacent matrix of corresponding nodes according to a three-phase load sequence of the low-voltage distribution transformer and constructing a space-time diagram data and diagram data sample facing the load prediction of the low-voltage distribution station area;
constructing corresponding multi-source condition data characteristics based on multi-source condition data, wherein the multi-source condition data comprise week types, time information, holidays, temperatures, humidity, wind speeds and air pressures;
the training module is used for training a condition self-adaptive space-time synchronization map convolutional neural network (CASSTGCN) model according to the constructed space-time map data and the multi-source condition data characteristics and excavating a load mode of a low-voltage distribution substation area;
the prediction module is used for applying the trained CASSTGCN model to predict the low-voltage distribution transformer three-phase load and respectively obtaining the load prediction values of the distribution transformer three phases at the next moment;
and the aggregation module is used for aggregating the load predicted values of the distribution transformer three phases to obtain the total load predicted value of the low-voltage distribution substation area at the next moment.
Preferably, the building module builds a characteristic sequence and an adjacency matrix of corresponding nodes according to a three-phase load sequence of the low-voltage distribution transformer, and building a space-time diagram data and diagram data sample facing load prediction of the low-voltage distribution substation area includes:
taking the three-phase load sequences as the characteristic sequences of each node of the graph structure data, and calculating the correlation coefficient among the phase load sequences;
constructing an adjacency matrix A according to the correlation coefficients, wherein,
The graph data samples are:
wherein,
is an input of the sample, and the sample is input,
in order to output the samples, the samples are,
a graph signal matrix formed of node features for time t, A ij Is the element adjacent to the ith row and the jth column of the matrix A, xi is the threshold value, rho ij Representing a load sequence X i And X j Pearson's correlation Coefficient of (COV) (X) i ,X j ) Represents X i And X j The covariance of (a) is determined,
and
respectively represent X i And X j Standard deviation of (2).
Preferably, the building module builds corresponding multi-source condition data characteristics based on the multi-source condition data, and the building module includes:
respectively constructing a week type sequence W, a time index sequence D, a holiday sign sequence H, a temperature characteristic sequence E, a humidity characteristic sequence M, a wind speed characteristic sequence P and an air pressure characteristic sequence Q, wherein at the time t, a condition data characteristic sequence L consists of the 7 sequences and is represented as follows:
L＝[W,D,H,E,M,P,Q]。
preferably, the training module training the condition adaptive space-time synchronization map convolutional neural network CASSGCN model comprises:
setting parameters of the CASSGCN model, wherein the parameters comprise: the method comprises the steps of historical load sequence input length, condition data characteristic dimension, input transformation layer dimension, space-time embedding layer dimension, graph volume layer number, sliding window length, graph volume layer activation function, output mapping layer dimension, output mapping layer activation function, learning rate, loss function, learning attenuation rate, batch scale and training period.
Preferably, the aggregating module aggregates the load predicted values of the distribution transformer three phases, and obtaining the total load predicted value of the low-voltage distribution substation area at the next time includes:
the calculation formula of the total load predicted value is as follows:
in the formula, P next The total load of the low-voltage distribution area is predicted,
respectively predicting the load of the A, B and C three phases of the low-voltage distribution transformer at the next time in the future according to the CATSGCN modelThe value is obtained.
Has the beneficial effects that: according to the method, the time correlation of the station area historical load sequence and the space correlation of the user electricity utilization behaviors in the station area can be mined by using the graph neural network, and the influence of multi-source external factors such as week types, time information and weather on the station area load can be mined, so that the short-term load prediction precision of the low-voltage distribution station area is remarkably improved. Therefore, the method is suitable for short-term load prediction of the low-voltage distribution substation area with strong space-time correlation, and under the background of implementation unit system, fine planning and operation maintenance of the power distribution network, the load prediction is carried out by taking the low-voltage distribution substation area as a research object, so that the method has important application value and popularization prospect in service scenes such as gridding load development trend and planning demand analysis, power distribution network weak link identification and power system operation mode adjustment.
The features and advantages of the present invention will become apparent by reference to the following drawings and detailed description of specific embodiments of the invention.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are included to illustrate an exemplary embodiment of the invention and not to limit the invention. In the drawings:
FIG. 1 is a flow chart of a method for forecasting short-term load of a platform level power;
FIG. 2 is a schematic diagram of a adjacency matrix and space-time diagram structure;
FIG. 3 is a schematic diagram of a time-space diagram data structure oriented to low-voltage distribution station load prediction;
FIG. 4 is a schematic diagram of a convolutional neural network structure of a condition-adaptive space-time synchronization map;
fig. 5 is a schematic diagram of a platform level short term load forecasting system.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings in conjunction with embodiments. It should be noted that the embodiments and features of the embodiments of the present invention may be combined with each other without conflict.
The following detailed description is exemplary in nature and is intended to provide further explanation of the invention as claimed. Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention.
Example 1
Fig. 1 is a flow chart of a method for predicting a short-term load of a platform level power. As shown in fig. 1, the present embodiment provides a method for predicting short-term load of power at a transformer area level, the method including the following steps:
s1: according to the three-phase load sequence of the low-voltage distribution transformer, a characteristic sequence and an adjacent matrix of corresponding nodes are established, and a time-space diagram data and diagram data sample facing to the load prediction of the low-voltage distribution transformer area are established.
Preferably, the establishing of the characteristic sequence and the adjacency matrix of the corresponding node according to the three-phase load sequence of the low-voltage distribution transformer and the constructing of the time-space diagram data and diagram data sample facing the load prediction of the low-voltage distribution substation area include:
taking the three-phase load sequences as the characteristic sequences of each node of the graph structure data, and calculating the correlation coefficient among the phase load sequences;
constructing an adjacency matrix A according to the correlation coefficients, wherein,
The graph data samples are:
wherein,
in order to input the samples, the method comprises the following steps of,
in order to output the samples, the samples are,
a graph signal matrix formed of node features for time t, A ij Is the element adjacent to the ith row and the jth column of the matrix A, xi is the threshold value, rho ij Representing a load sequence X i And X j Pearson's correlation Coefficient of (COV) (X) i ,X j ) Represents X i And X j The covariance of (a) of (b),
and
respectively represent X i And X j Standard deviation of (2).
Specifically, referring to fig. 2, first, the three-phase load sequence is taken as a feature sequence of each node of the graph structure data, and then the correlation coefficient between the load sequences of the phases is calculated as follows:
in the formula: rho ij Representing a load sequence X i And X j Pearson's correlation coefficient, COV (X) i ,X j ) Represents X i And X j The covariance of (a) of (b),
and
respectively represent X i And X j Standard deviation of (2).
The first adjacency matrix construction method is as follows: if the correlation coefficient between the corresponding historical load sequences of the two nodes is not smaller than a certain threshold value, the two nodes are considered to have a connection relation, and the corresponding position element of the adjacent matrix A is set to be 1; otherwise, it is considered that there is no connection relationship between the two nodes, and the corresponding position element of the adjacency matrix a is set to 0. The specific calculation formula is as follows:
in the formula, A ij Is the element adjacent to row i and column j of matrix a, ξ is the threshold.
The second adjacency matrix construction method is as follows: the correlation coefficient between the respective historical load sequences of two nodes is used as the corresponding element of the adjacency matrix. The specific calculation formula is as follows:
A ij ＝ρ ij (3)
(2) And constructing the space-time diagram data facing the load prediction of the low-voltage distribution station area based on the three-phase load sequence and the corresponding adjacent matrix.
(3) Construction of graph data sample for low-voltage distribution station load prediction
Wherein,
is an input of the sample, and the sample is input,
is the output of the sample, and the output of the sample,
a graph signal matrix formed by node characteristics at time T is shown, A is an adjacent matrix, and T is the length of a history sequence. The time-space diagram data structure facing the low-voltage distribution station load prediction is shown in fig. 3, wherein each node is characterized by a load value at a corresponding moment.
S2: and constructing corresponding multi-source condition data characteristics based on multi-source condition data, wherein the multi-source condition data comprises week types, time information, holidays, temperature, humidity, wind speed and air pressure.
Preferably, the constructing corresponding multi-source condition data characteristics based on the multi-source condition data comprises:
respectively constructing a week type sequence W, a time index sequence D, a holiday sign sequence H, a temperature characteristic sequence E, a humidity characteristic sequence M, a wind speed characteristic sequence P and an air pressure characteristic sequence Q, wherein for a time t, a condition data characteristic sequence L consists of the 7 sequences and is represented as follows:
L＝[W,D,H,E,M,P,Q]。
specifically, for a graph data sample at the time t, corresponding condition data features are constructed, which include 7 multi-source data features such as week type, time information, holiday information, temperature, humidity, wind speed, and air pressure, and are specifically expressed as follows:
1) For the past T times, the week type sequence W is represented as:
W＝[w t-T+1 ,...,w t-1 ,w t ] (4)
wherein, w t ∈[1，7]The week type at time t;
2) For the past T time instants, the time-instant index sequence D is represented as:
D＝[d t-T+1 ,...,d t-1 ,d t ] (5)
wherein d is t ∈[1,F]F is the sampling frequency of the load data, and is generally 24, 48 or 96;
3) For the past T moments, the holiday flag sequence H is represented as:
H＝[h t-T+1 ,...,h t-1 ,h t ] (6)
wherein h is t For the holiday flag at time t, the value is 1 or 2, with 1 representing a non-holiday and 2 representing a holiday.
4) For the past T moments, the temperature signature sequence E is represented as:
E＝[e t-T+1 ,...,e t-1 ,e t ] (7)
wherein,e t is the temperature at time t;
5) For the past T moments, the humidity signature sequence M is represented as:
M＝[m t-T+1 ,...,m t-1 ,m t ] (8)
wherein m is t Humidity at time t;
6) For the past T moments, the wind speed signature sequence P is represented as:
P＝[p t-T+1 ,...,p t-1 ,p t ] (9)
wherein p is t Is the wind speed at time t;
7) For the past T time instants, the barometric pressure profile Q is represented as:
Q＝[q t-T+1 ,...,q t-1 ,q t ] (10)
wherein q is t Is the air pressure at time t;
thus, for time t, the conditional data signature sequence L consists of the above 7 sequences, denoted as:
L＝[W,D,H,E,M,P,Q] (11)
s3: and training a condition self-adaptive space-time synchronization map convolutional neural network (CASSTGCN) model according to the constructed space-time map data and the multi-source condition data characteristics, and excavating a load mode of a low-voltage distribution substation area.
Preferably, the training condition adaptive space-time synchronization map convolutional neural network CASSGCN model comprises:
setting parameters of the CASSGCN model, wherein the parameters comprise: the method comprises the steps of historical load sequence input length, condition data characteristic dimension, input transformation layer dimension, space-time embedding layer dimension, graph volume layer number, sliding window length, graph volume layer activation function, output mapping layer dimension, output mapping layer activation function, learning rate, loss function, learning attenuation rate, batch scale and training period.
Specifically, the CASSGCN model is trained according to the graph data samples constructed in the steps 1 and 2 and corresponding condition data characteristics. The structure of the CASSGCN model is shown in FIG. 4, and the corresponding parameter settings are shown in Table 1.
TABLE 1 CASSGCN model parameter Table
S4: and predicting the low-voltage distribution transformation three-phase load by applying the trained CASSTGCN model to respectively obtain the load predicted values of the distribution transformation three phases at the next moment.
Specifically, the trained CASSTGCN model is applied to predict the low-voltage distribution transformation three-phase load to obtain
Wherein,
and load predicted values of the three phases A, B and C of the low-voltage distribution transformer at the next moment in the future are respectively.
S5: and aggregating the load predicted values of the three phases of the distribution transformer to obtain a total load predicted value of the low-voltage distribution substation area at the next moment.
Preferably, the aggregating the load predicted values of the distribution transformation three phases to obtain a total load predicted value of the low-voltage distribution substation area at the next time includes:
the calculation formula of the total load predicted value is as follows:
in the formula, P next The total load prediction value of the low-voltage distribution station area is obtained,
and the load predicted values of the three phases A, B and C of the low-voltage distribution transformer at the next future moment are predicted according to the CASSTGCN model.
Specifically, the predicted load value of the low-voltage distribution transformer three phases is aggregated, so as to obtain the predicted total load value of the low-voltage distribution transformer area at the next moment in the future, and a specific calculation formula is as follows:
in the formula, P next And predicting the total load of the low-voltage distribution area.
In the embodiment, the graph neural network is utilized, so that the time correlation of the station area historical load sequence and the space correlation between the user electricity utilization behaviors in the station area can be mined, and the influence of multi-source external factors such as week types, time information and weather on the station area load can be mined, so that the short-term load prediction precision of the low-voltage distribution station area is remarkably improved. Therefore, the method is suitable for short-term load prediction of the low-voltage distribution area with strong space-time correlation, load prediction is carried out by taking the low-voltage distribution area as a research object under the background of power distribution network implementation unit system, fine planning and operation maintenance, and the method has important application value and popularization prospect in service scenes such as gridding load development trend, planning demand analysis, power distribution network weak link identification and power system operation mode adjustment.
Example 2
Fig. 5 is a schematic diagram of a platform level short term load forecasting system. As shown in fig. 5, the present embodiment provides a platform level power short-term load prediction system, which includes:
the building module 501 is used for building a characteristic sequence and an adjacent matrix of corresponding nodes according to a three-phase load sequence of the low-voltage distribution transformer, and building a time-space diagram data and a diagram data sample facing the load prediction of the low-voltage distribution transformer area;
constructing corresponding multi-source condition data characteristics based on multi-source condition data, wherein the multi-source condition data comprise week types, time information, holidays, temperatures, humidity, wind speeds and air pressures;
the training module 502 is used for training a condition adaptive space-time synchronization map convolutional neural network (CASSTGCN) model according to the constructed space-time map data and the multi-source condition data characteristics, and excavating a load mode of a low-voltage distribution substation area;
the prediction module 503 is configured to apply the trained casstsgcn model to predict the low-voltage distribution transformer three-phase load, and obtain load prediction values of the distribution transformer three phases at the next moment respectively;
and an aggregation module 504, configured to aggregate the load predicted values of the distribution transformer three phases to obtain a total load predicted value of the low-voltage distribution substation area at the next time.
Preferably, the constructing module 501 establishes a feature sequence and an adjacency matrix of corresponding nodes according to a three-phase load sequence of the low-voltage distribution transformer, and constructing a time-space diagram data and a diagram data sample facing to the load prediction of the low-voltage distribution substation area includes:
taking the three-phase load sequences as the characteristic sequences of each node of the graph structure data, and calculating the correlation coefficient among the phase load sequences;
constructing an adjacency matrix A according to the correlation coefficients, wherein,
The graph data samples are:
wherein,
is an input of the sample, and the sample is input,
in order to output the samples, the samples are,
a graph signal matrix formed of node features for time t, A ij Is the element adjacent to the ith row and jth column of matrix A, xi is the threshold value, rho ij Representing a load sequence X i And X j Pearson's correlation Coefficient of (COV) (X) i ,X j ) Represents X i And X j The covariance of (a) is determined,
and
respectively represent X i And X j Standard deviation of (d).
Preferably, the constructing module 501 constructs, based on the multi-source condition data, a corresponding feature of the multi-source condition data, including:
respectively constructing a week type sequence W, a time index sequence D, a holiday sign sequence H, a temperature characteristic sequence E, a humidity characteristic sequence M, a wind speed characteristic sequence P and an air pressure characteristic sequence Q, wherein at the time t, a condition data characteristic sequence L consists of the 7 sequences and is represented as follows:
L＝[W,D,H,E,M,P,Q]。
preferably, the training module 502 trains the condition-adaptive spatio-temporal synchronization map convolutional neural network catsgcn model to include:
setting parameters of the CASSGCN model, wherein the parameters comprise: the method comprises the steps of historical load sequence input length, condition data characteristic dimension, input transformation layer dimension, space-time embedding layer dimension, graph volume layer number, sliding window length, graph volume layer activation function, output mapping layer dimension, output mapping layer activation function, learning rate, loss function, learning attenuation rate, batch scale and training period.
Preferably, the aggregating module 504 aggregates the load prediction values of the distribution transformer three phases, and obtaining the total load prediction value of the low-voltage distribution substation at the next time includes:
the calculation formula of the total load predicted value is as follows:
in the formula, P next The total load of the low-voltage distribution area is predicted,
and the load predicted values of the three phases A, B and C of the low-voltage distribution transformer at the next future moment are predicted according to the CASSTGCN model.
The specific implementation process of the functions implemented by each module in this embodiment 2 is the same as the implementation process of each step in embodiment 1, and is not described herein again.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting the same, and although the present invention is described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.
Claims (10)
1. A method for predicting a short-term load of a platform-level power, the method comprising the steps of:
according to a three-phase load sequence of the low-voltage distribution transformer, establishing a characteristic sequence and an adjacent matrix of corresponding nodes, and constructing a space-time diagram data and diagram data sample facing to low-voltage distribution station load prediction;
constructing corresponding multi-source condition data characteristics based on multi-source condition data, wherein the multi-source condition data comprise week types, time information, holidays, temperatures, humidity, wind speeds and air pressures;
training a condition self-adaptive space-time synchronization map convolutional neural network (CATSGCN) model according to the constructed space-time map data and the multi-source condition data characteristics, and excavating a load mode of a low-voltage distribution substation area;
predicting the low-voltage distribution transformer three-phase load by applying the trained CASSTGCN model to respectively obtain the load prediction values of the distribution transformer three phases at the next moment;
and aggregating the load predicted values of the distribution and transformation three phases to obtain the total load predicted value of the low-voltage distribution substation at the next moment.
2. The method of claim 1, wherein the characteristic sequence and the adjacency matrix of the corresponding nodes are established according to the three-phase load sequence of the low-voltage distribution transformer, and the constructing of the space-time diagram data and the diagram data samples facing the low-voltage distribution station load prediction comprises the following steps:
taking the three-phase load sequences as the characteristic sequences of each node of the graph structure data, and calculating the correlation coefficient among the phase load sequences;
constructing an adjacency matrix A according to the correlation coefficients, wherein,
The graph data samples are:
wherein,
in order to input the samples, the method comprises the following steps of,
in order to output the samples, the samples are,
a graph signal matrix formed of node features for time t, A ij Is the element adjacent to the ith row and the jth column of the matrix A, xi is the threshold value, rho ij Representing a load sequence X i And X j Pearson's correlation Coefficient of (COV) (X) i ,X j ) Represents X i And X j The covariance of (a) of (b),
and
respectively represent X i And X j Standard deviation of (2).
3. The method of claim 2, wherein constructing the corresponding multi-source condition data feature based on the multi-source condition data comprises:
respectively constructing a week type sequence W, a time index sequence D, a holiday sign sequence H, a temperature characteristic sequence E, a humidity characteristic sequence M, a wind speed characteristic sequence P and an air pressure characteristic sequence Q, wherein at the time t, a condition data characteristic sequence L consists of the 7 sequences and is represented as follows:
L＝[W,D,H,E,M,P,Q]。
4. the method of claim 3, wherein the training condition adaptive space-time synchronization map convolutional neural network (CASSGCN) model comprises:
setting parameters of the CASSGCN model, wherein the parameters comprise: the method comprises the steps of historical load sequence input length, condition data characteristic dimension, input transformation layer dimension, space-time embedding layer dimension, graph volume layer number, sliding window length, graph volume layer activation function, output mapping layer dimension, output mapping layer activation function, learning rate, loss function, learning attenuation rate, batch scale and training period.
5. The method of claim 4, wherein aggregating the load prediction values of the distribution transformation three phases to obtain a total load prediction value of a low-voltage distribution substation at a next time comprises:
the calculation formula of the total load predicted value is as follows:
in the formula, P next The total load prediction value of the low-voltage distribution station area is obtained,
and the load prediction values of the A, B and C three phases of the low-voltage distribution transformer at the next time are predicted according to the CASSGCN model.
6. A district-level short-term load forecasting system, the system comprising:
the construction module is used for establishing a characteristic sequence and an adjacent matrix of corresponding nodes according to a three-phase load sequence of the low-voltage distribution transformer and constructing a space-time diagram data and diagram data sample facing the load prediction of the low-voltage distribution station area;
constructing corresponding multi-source condition data characteristics based on multi-source condition data, wherein the multi-source condition data comprise week types, time information, holidays, temperatures, humidity, wind speeds and air pressures;
the training module is used for training a condition self-adaptive space-time synchronization map convolutional neural network (CASSGCN) model according to the constructed space-time map data and the multi-source condition data characteristics and excavating a load mode of a low-voltage power distribution station area;
the prediction module is used for applying the trained CASSTGCN model to predict the low-voltage distribution transformer three-phase load and respectively obtaining the load prediction values of the distribution transformer three phases at the next moment;
and the aggregation module is used for aggregating the load predicted values of the distribution transformer three phases to obtain the total load predicted value of the low-voltage distribution substation area at the next moment.
7. The system of claim 6, wherein the construction module establishes a characteristic sequence and an adjacency matrix of corresponding nodes according to a three-phase load sequence of the low-voltage distribution transformer, and the construction of the spatio-temporal graph data and the graph data samples facing the low-voltage distribution station load prediction comprises:
taking the three-phase load sequences as the characteristic sequences of each node of the graph structure data, and calculating the correlation coefficient among the phase load sequences;
constructing an adjacency matrix A according to the correlation coefficients, wherein,
The graph data samples are:
wherein,
is an input of the sample, and the sample is input,
in order to output the samples, the samples are,
a graph signal matrix formed of node features for time t, A ij Is the element adjacent to the ith row and the jth column of the matrix A, xi is the threshold value, rho ij Representing a load sequence X i And X j Pearson's correlation coefficient, COV (X) i ,X j ) Represents X i And X j The covariance of (a) of (b),
and
respectively represent X i And X j Standard deviation of (2).
8. The system of claim 7, wherein the build module builds, based on the multi-source condition data, the corresponding multi-source condition data features comprising:
respectively constructing a week type sequence W, a time index sequence D, a holiday sign sequence H, a temperature characteristic sequence E, a humidity characteristic sequence M, a wind speed characteristic sequence P and an air pressure characteristic sequence Q, wherein at the time t, a condition data characteristic sequence L consists of the 7 sequences and is represented as follows:
L＝[W,D,H,E,M,P,Q]。
9. the system of claim 8, wherein the training module trains the condition-adaptive spatio-temporal synchronization map convolutional neural network (CASSGCN) model to include:
setting parameters of the CASSGCN model, wherein the parameters comprise: the method comprises the steps of historical load sequence input length, condition data characteristic dimension, input transformation layer dimension, space-time embedding layer dimension, graph volume layer number, sliding window length, graph volume layer activation function, output mapping layer dimension, output mapping layer activation function, learning rate, loss function, learning attenuation rate, batch scale and training period.
10. The system of claim 9, wherein the aggregation module aggregates load forecasts of the distribution transformation three phases to obtain a total load forecast for a low voltage distribution substation at a next time comprises:
the calculation formula of the total load predicted value is as follows:
in the formula, P next The total load of the low-voltage distribution area is predicted,
and the load prediction values of the A, B and C three phases of the low-voltage distribution transformer at the next time are predicted according to the CASSGCN model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211527225.3A CN115730740A (en) | 2022-11-30 | 2022-11-30 | Transformer area level power short-term load prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211527225.3A CN115730740A (en) | 2022-11-30 | 2022-11-30 | Transformer area level power short-term load prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115730740A true CN115730740A (en) | 2023-03-03 |
Family
ID=85299669
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211527225.3A Pending CN115730740A (en) | 2022-11-30 | 2022-11-30 | Transformer area level power short-term load prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115730740A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118316037A (en) * | 2024-06-07 | 2024-07-09 | 北京智芯微电子科技有限公司 | Intelligent transformer area power load prediction method, device and storage medium |
-
2022
- 2022-11-30 CN CN202211527225.3A patent/CN115730740A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118316037A (en) * | 2024-06-07 | 2024-07-09 | 北京智芯微电子科技有限公司 | Intelligent transformer area power load prediction method, device and storage medium |
| CN118316037B (en) * | 2024-06-07 | 2024-08-13 | 北京智芯微电子科技有限公司 | Intelligent transformer area power load prediction method, device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113962364B (en) | Multi-factor power load prediction method based on deep learning | |
| Khan et al. | Genetic algorithm based optimized feature engineering and hybrid machine learning for effective energy consumption prediction | |
| CN116646933A (en) | Big data-based power load scheduling method and system | |
| CN111552923A (en) | Load prediction method and load prediction system based on universal distribution | |
| CN108053082B (en) | Power grid medium and long term load prediction method based on temperature interval decomposition | |
| CN112801342A (en) | Adaptive runoff forecasting method based on rainfall runoff similarity | |
| CN111680841B (en) | Short-term load prediction method, system and terminal equipment based on principal component analysis | |
| CN106910144B (en) | Large building time-by-time energy consumption online prediction method based on time-division actual energy consumption coefficient | |
| CN112149890A (en) | Comprehensive energy load prediction method and system based on user energy label | |
| CN112508306A (en) | Self-adaptive method and system for power production configuration | |
| CN114372360A (en) | Method, terminal and storage medium for power load prediction | |
| Dong et al. | Short-term building cooling load prediction model based on DwdAdam-ILSTM algorithm: A case study of a commercial building | |
| Porteiro et al. | Short term load forecasting of industrial electricity using machine learning | |
| CN115481918A (en) | Active sensing and predictive analysis system for unit state based on source network load storage | |
| Yu et al. | Short-term cooling and heating loads forecasting of building district energy system based on data-driven models | |
| CN109657874A (en) | A kind of electric power Mid-long term load forecasting method based on long memory models in short-term | |
| CN108346009A (en) | A kind of power generation configuration method and device based on user model self study | |
| CN115730740A (en) | Transformer area level power short-term load prediction method and system | |
| Qiao et al. | Predicting building energy consumption based on meteorological data | |
| CN111160993A (en) | User practical and stage release coefficient analysis method based on user load application | |
| Zheng et al. | Short-term energy consumption prediction of electric vehicle charging station using attentional feature engineering and multi-sequence stacked Gated Recurrent Unit | |
| CN117932470A (en) | Moon runoff forecasting method based on dynamic Bayesian network | |
| Zou et al. | Predicting the electric power consumption of office buildings based on dynamic and static hybrid data analysis | |
| CN117200223A (en) | Day-ahead power load prediction method and device | |
| CN115130788B (en) | Big data analysis-based power utilization prediction system |
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 |