CN116108960A - Training method and device for multi-type energy demand prediction model - Google Patents
Training method and device for multi-type energy demand prediction model Download PDFInfo
- Publication number
- CN116108960A CN116108960A CN202211534854.9A CN202211534854A CN116108960A CN 116108960 A CN116108960 A CN 116108960A CN 202211534854 A CN202211534854 A CN 202211534854A CN 116108960 A CN116108960 A CN 116108960A
- Authority
- CN
- China
- Prior art keywords
- energy demand
- data
- encoder
- type energy
- self
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06315—Needs-based resource requirements planning or analysis
-
- 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)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Educational Administration (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a training method and a training device for a multi-type energy demand prediction model, wherein the method comprises the following steps: acquiring a first energy demand influence data set and a corresponding multi-type energy demand data set; performing feature extraction on the first energy demand influence data set by using a trained encoder of the self-encoder to obtain a second energy demand influence data set; and training the time sequence prediction model to be trained by using the second energy demand influence data set and the multi-type energy demand data set to obtain the multi-type energy demand prediction model. The method and the device for training the multi-type energy demand prediction model can effectively improve the training efficiency of the model and the generalization capability of the model.
Description
Technical Field
The invention relates to the technical field of energy demand, in particular to a training method and device for a multi-type energy demand prediction model.
Background
The energy demand prediction has important significance for the development of the energy industry and the supply service of the energy demand, at present, most of energy demand prediction methods are simple, the training of the adopted model is low-efficiency, the generalization capability is poor, and therefore the finally predicted energy demand data is inaccurate.
Disclosure of Invention
The purpose of the invention is that: the multi-type energy demand prediction model training method, the multi-type energy demand prediction device, the computer equipment and the storage medium can effectively improve the training efficiency of the model and the generalization capability of the model, so that the finally predicted energy demand data can be more accurate.
To achieve the above object, in a first aspect, the present invention provides a method for training a multi-type energy demand prediction model, including:
acquiring a first energy demand influence data set and a corresponding multi-type energy demand data set;
performing feature extraction on the first energy demand influence data set by using a trained encoder of the self-encoder to obtain a second energy demand influence data set;
and training the time sequence prediction model to be trained by using the second energy demand influence data set and the multi-type energy demand data set to obtain the multi-type energy demand prediction model.
In a preferred embodiment of the present invention, after the obtaining the first energy requirement influence dataset and the corresponding multi-type energy requirement dataset, the feature extraction is performed on the first energy requirement influence dataset by using the trained encoder of the self-encoder, and before obtaining the second energy requirement influence dataset, the method further includes:
training a self-encoder to be trained using the first energy demand impact dataset, the self-encoder comprising an encoder and a decoder;
-disassembling the decoder in the self-encoder.
In a preferred embodiment of the present invention, the training the self-encoder to be trained using the first energy requirement influence data set includes:
respectively inputting the data in the first energy requirement influence data set into a self-encoder to be trained for iterative training;
in each of the iterative training steps,
the encoder of the self-encoder performs feature extraction on input data to obtain feature vectors;
the decoder of the self-encoder reconstructs the feature vector to obtain reconstructed data;
and updating the parameters of the self-encoder by using a predetermined loss function, wherein the predetermined loss function is defined according to the distance between the reconstruction data and the input data.
In a preferred embodiment of the present invention, the updating the parameter of the self-encoder with a predetermined loss function includes:
and updating the parameters of the self-encoder in the direction of the reduction of the preset loss function by using a gradient descent method.
In a preferred embodiment of the present invention, the data in the first energy demand impact data set includes meteorological data, time data, power energy data, grid scheduling data, socioeconomic data, and carbon emission data.
In a second aspect, the present invention provides a method for predicting multi-type energy demand, comprising:
acquiring energy demand influence data to be predicted;
performing feature extraction on the energy demand influence data to be predicted by using a trained encoder of the self-encoder to obtain target energy demand influence data;
and inputting the target energy demand influence data into a multi-type energy demand prediction model, and predicting to obtain target multi-type energy demand data.
In a third aspect, the present invention provides a multi-type energy demand prediction model training apparatus, comprising:
the acquisition module is used for acquiring the first energy demand influence data set and the corresponding multi-type energy demand data set;
the feature extraction module is used for carrying out feature extraction on the first energy demand influence data set by utilizing the trained encoder of the self-encoder to obtain a second energy demand influence data set;
and the training module is used for training the time sequence prediction model to be trained by utilizing the second energy demand influence data set and the multi-type energy demand data set to obtain the multi-type energy demand prediction model.
In a fourth aspect, the present invention provides a multi-type energy demand prediction apparatus comprising:
the acquisition module is used for acquiring the energy demand influence data to be predicted;
the feature extraction module is used for extracting features of the energy demand influence data to be predicted by utilizing the trained encoder of the self-encoder to obtain target energy demand influence data;
and the energy demand prediction module is used for inputting the target energy demand influence data into a multi-type energy demand prediction model to predict and obtain target multi-type energy demand data.
In a fifth aspect, the present invention provides a computer device, including a memory and a processor, where the memory is configured to store a computer program, and the processor is configured to execute the computer program to cause the computer device to perform the above-described multi-type energy demand prediction model training method, or the above-described multi-type energy demand prediction method.
In a sixth aspect, the present invention provides a computer readable storage medium storing a computer program which when executed by a processor implements the above-described multi-type energy demand prediction model training method, or the above-described multi-type energy demand prediction method.
Compared with the prior art, the multi-type energy demand prediction model training method, the multi-type energy demand prediction device, the computer equipment and the storage medium have the beneficial effects that:
according to the embodiment of the invention, the first energy demand influence data set is obtained through feature extraction by the trained encoder of the self-encoder, the effective features of the data are extracted, and the second energy demand influence data set is obtained; and then training the time sequence prediction model to be trained by utilizing the second energy demand influence data set and the acquired multi-type energy demand data set, so that when the multi-type energy demand prediction model is obtained, the training efficiency of the multi-type energy demand prediction model can be effectively improved, the generalization capability of the model is improved, the finally predicted energy demand data can be more accurate, the energy demand influence data and the multi-type energy demand data are adopted during model training, and the multi-type energy demand data can be finally predicted, so that the effect of energy demand prediction is greatly improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments of the present invention will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and should not be considered as limiting the scope, and other related drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a training method for a multi-type energy demand prediction model according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a self-encoder according to a first embodiment of the present invention;
FIG. 3 is a block diagram of a training device for a multi-type energy demand prediction model according to a second embodiment of the present invention;
FIG. 4 is a flow chart of a method for predicting multi-type energy demand according to a third embodiment of the present invention;
FIG. 5 is a block diagram of a multi-type energy demand prediction apparatus according to a fourth embodiment of the present invention;
fig. 6 is a schematic diagram of an internal structure of a computer device according to a fifth embodiment of the present invention.
Detailed Description
The following describes in further detail the embodiments of the present invention with reference to the drawings and examples. The following examples are illustrative of the invention and are not intended to limit the scope of the invention.
At present, most of energy demand prediction methods are simple, the training of the adopted model is low-efficiency, and the generalization capability is poor, so that the finally predicted energy demand data is inaccurate.
Aiming at the problems in the prior art, the embodiment of the invention provides a multi-type energy demand prediction model training method, a multi-type energy demand prediction device, a multi-type energy demand prediction computer device and a multi-type energy demand prediction storage medium, which can effectively improve the training efficiency of the multi-type energy demand prediction model and the generalization capability of the model, so that the finally predicted energy demand data is more accurate, and the energy demand influence data and the multi-type energy demand data are adopted in the model training, so that the multi-type energy demand data can be finally predicted, and the energy demand prediction effect is greatly improved.
Example 1
Referring to fig. 1, fig. 1 is a schematic flow chart of a training method of a multi-type energy demand prediction model according to an embodiment of the present invention.
The multi-type energy demand prediction model training method in the embodiment of the invention can be applied to computer equipment such as a server.
In one embodiment, the invention provides a multi-type energy demand prediction model training method, which comprises the following steps:
step S110, a first energy requirement influence dataset and a corresponding multi-type energy requirement dataset are obtained.
In one embodiment, the data in the obtained first energy demand impact data set may include data such as meteorological data, time data, electrical energy data, grid scheduling data, socioeconomic data, carbon emission data, etc.; optionally, each data in the first energy requirement influence data set is data corresponding to the current time when the data is acquired, that is, data such as meteorological data, time data, electric power energy data, power grid dispatching data, social economic data, carbon emission data and the like is data corresponding to the current time when each data is acquired;
wherein, the meteorological data can comprise at least one of data such as the highest air temperature, the lowest air temperature, the average precipitation, the average relative humidity, the wind speed, the average air pressure and the like; the electric power energy data can comprise at least one of data such as grid-connected installed capacity of coal engines, nuclear power, gas power, water power and the like, output data, wind-light power generation utilization hours, energy consumption data (such as unit GDP power consumption, unit total production value energy consumption, first industry energy consumption, per-person coal consumption and the like) and the like; the socioeconomic data may include at least one of GDP, average income, urbanization rate, first industry growth rate, second industry growth rate, third industry growth rate, population, two-three industry ratio, sub-industry GDP, etc.; the carbon emission data may include data such as carbon dioxide emissions per unit total production.
In one embodiment, the data in the acquired multi-type energy demand dataset may include energy demand data such as power demand data, coal demand data, oil demand data, natural gas demand data, and the like; the data in the multi-type energy demand data set corresponds to the data in the first energy demand impact data set.
In one embodiment, the first energy demand impact dataset and the corresponding multi-type energy demand dataset may be time ordered and normalized.
In one embodiment, the computer device may obtain the first energy requirement influence data set and the corresponding multi-type energy requirement data set by using the energy requirement influence data and the corresponding multi-type energy requirement data input by the staff.
In one embodiment, a first energy demand impact dataset is used as the input dataset and a corresponding multi-type energy demand dataset is used as the output dataset.
In one embodiment, the data in the first energy demand influence data set includes more and more data types, so that training of the time sequence prediction model can be facilitated, and the obtained multi-type energy demand prediction model has the capability of predicting the energy demand data more accurately.
Step S120, the first energy requirement influence data set is extracted by utilizing the trained encoder of the self-encoder to obtain a second energy requirement influence data set.
In one embodiment, before the step S120, the method for training a multi-type energy demand prediction model according to the present invention may further include the following steps:
training a self-encoder to be trained using the first energy demand impact dataset, the self-encoder comprising an encoder and a decoder;
the decoder in the self-encoder is disassembled.
In this embodiment, the self-encoder is constructed by using a convolutional neural network, and is divided into an encoder and a decoder, and a schematic diagram of the self-encoder in this embodiment is shown in fig. 2, where the Maxpooling layer and the Dropout layer after the Conv1D layer are omitted in fig. 2;
for the encoder in the self-encoder, the function of the self-encoder is to extract the data characteristics, the function of the decoder in the self-encoder is to reconstruct the data in a reducing way, and the reconstruction data is better as the reconstruction data is closer to the original data; through the mode, the self-encoder can learn the characteristics in the data better, has better capability of learning the data characteristics and extracting the characteristics, and disassembles the decoder in the self-encoder after reaching the capability, so that the self-encoder can be applied better.
In this embodiment, when the computer device trains the self-encoder to be trained using the first energy requirement influence data set, the computer device may:
respectively inputting data in the first energy requirement influence data set into a self-encoder to be trained for iterative training;
in each of the iterative training steps,
the method comprises the steps that an encoder of a self-encoder performs feature extraction on input data to obtain feature vectors;
reconstructing the feature vector from a decoder of the encoder to obtain reconstructed data;
updating parameters of the self-encoder by using a predetermined loss function, wherein the predetermined loss function is defined according to the distance between the reconstruction data and the input data;
specifically, for the above process, reference may be made to fig. 2, wherein the first energy requirement influence data set is referred to as the multi-energy input sample data set in fig. 2, and the reconstructed data set is referred to as the reconstructed multi-energy input sample data set in fig. 2;
optionally, before training the self-encoder to be trained using the first energy requirement influence data set, the first energy requirement influence data set may be divided into a training set, a verification set and a test set according to a predetermined division ratio; in the training process of the self-encoder, each epoch (which means that all data are sent into a network and the forward calculation and the backward propagation processes are completed once) is verified by using a verification set, and the performance and the fitting degree of the self-encoder are evaluated by using a test set after the training is finished;
when the encoder of the self-encoder extracts the characteristics of the input data to obtain the characteristic vector, the input data can be subjected to characteristic extraction and compressed to the characteristic vector, and the characteristic vector can be regarded as the representation of the input data;
the predetermined loss function may be expressed as follows:
wherein X is i Representing the ith input data, χ i Representing reconstruction data corresponding to the ith input data, N representing the number of input data;
in the present embodiment, when the parameter of the self-encoder is updated by using a predetermined loss function, the parameter of the self-encoder may be updated in a direction in which the predetermined loss function decreases by using a gradient descent method; optionally, the parameters of the self-encoder can be initialized first, then the self-encoder is trained by using L as a loss function by using a gradient descent method, so that the parameters in the self-encoder are continuously and iteratively updated towards the descending direction of the loss function L, and the distance between the reconstructed data and the input data is more and more similar.
In this embodiment, the self-encoder can be better trained in the above manner, the training time and times are shortened, the training effect is improved, and the self-encoder can better learn the features in the data, and has better capability of learning the data features and extracting the features.
Step S130, training the time sequence prediction model to be trained by using the second energy demand influence data set and the multi-type energy demand data set to obtain the multi-type energy demand prediction model.
In one embodiment, the timing prediction model to be trained may be an LSTM (Long short-Term Memory network) based timing prediction model.
In this embodiment, the input data of the LSTM-based time series prediction model is sequence data, and the dimensions are (N, T, D), where N is the number of input data, T is the time length, and D is a single input data dimension (i.e., feature vector dimension); the LSTM-based time sequence prediction model can output multi-type energy demand data of future time (t+1, t+n) according to sequence data input by the current time and the past time (t-n, t-1), wherein the positive integer n is a step length; the dimension of single multi-type energy demand data at a certain time is d, d is the number of energy types, and if the dimension is a single dimension, the data of the single dimension is the demand data of the corresponding energy type;
optionally, when training the time sequence prediction model to be trained by using the second energy demand influence data set and the multi-type energy demand data set to obtain the multi-type energy demand prediction model, the second energy demand influence data set and the multi-type energy demand data set can be divided into a training set, a verification set and a test set according to a predetermined dividing proportion; in the training process of the time sequence prediction model, each epoch (which means that all data are sent into a network and the forward calculation and back propagation processes are completed once) is verified by applying a verification set until the model converges, and finally the performance and the fitting degree of the model are evaluated by using a test set;
optionally, when the time sequence prediction model is trained, the parameters of the time sequence prediction model can be initialized first, and then the time sequence prediction model is trained by using the average absolute error as a loss function through a gradient descent method.
In one embodiment, the time sequence prediction model can be trained better through the mode, so that the obtained multi-type energy demand prediction model has better generalization capability, the model has the capability of predicting energy demand data more accurately, and the prediction effect of the model is improved.
According to the multi-type energy demand prediction model training method, the first energy demand influence data set is obtained through feature extraction by the trained encoder of the self-encoder, the effective features of the data are extracted, and the second energy demand influence data set is obtained; and when the second energy demand influence data set and the acquired multi-type energy demand data set are utilized to train the time sequence prediction model to be trained, and the multi-type energy demand prediction model is obtained, the training efficiency of the multi-type energy demand prediction model can be effectively improved, the generalization capability of the model is improved, so that the model has the capability of predicting the energy demand data more accurately, the energy demand influence data and the multi-type energy demand data are adopted during model training, and the model can finally have the capability of predicting the multi-type energy demand data, so that the prediction effect of the model is greatly improved.
Example two
In order to execute the corresponding method of the above embodiment to achieve the corresponding functions and technical effects, a multi-type energy demand prediction model training device is provided below.
Referring to fig. 3, fig. 3 is a block diagram of a multi-type energy demand prediction model training device according to an embodiment of the present invention.
In one embodiment, the multi-type energy demand prediction model training device of the present invention comprises:
an obtaining module 210, configured to obtain a first energy requirement influence data set and a corresponding multi-type energy requirement data set;
the feature extraction module 220 is configured to perform feature extraction on the first energy requirement influence data set by using the trained encoder of the self-encoder, so as to obtain a second energy requirement influence data set;
the training module 230 is configured to train the time sequence prediction model to be trained by using the second energy demand influence data set and the multi-type energy demand data set, so as to obtain the multi-type energy demand prediction model.
According to the multi-type energy demand prediction model training device, the first energy demand influence data set is obtained through feature extraction by the trained encoder of the self-encoder, the effective features of the data are extracted, and the second energy demand influence data set is obtained; and when the second energy demand influence data set and the acquired multi-type energy demand data set are utilized to train the time sequence prediction model to be trained, and the multi-type energy demand prediction model is obtained, the training efficiency of the multi-type energy demand prediction model can be effectively improved, the generalization capability of the model is improved, so that the model has the capability of predicting the energy demand data more accurately, the energy demand influence data and the multi-type energy demand data are adopted during model training, and the model can finally have the capability of predicting the multi-type energy demand data, so that the prediction effect of the model is greatly improved.
In one embodiment, training module 230 may also be configured to:
training a self-encoder to be trained by using the first energy demand influence data set;
the multi-type energy demand prediction model training device of the invention can further comprise:
and the disassembly module is used for disassembling the encoder in the self-encoder.
In one embodiment, the training module 230 may be specifically configured to:
respectively inputting data in the first energy requirement influence data set into a self-encoder to be trained for iterative training;
in each of the iterative training steps,
the method comprises the steps that an encoder of a self-encoder performs feature extraction on input data to obtain feature vectors;
reconstructing the feature vector from a decoder of the encoder to obtain reconstructed data;
and updating the parameters of the self-encoder by using a predetermined loss function, wherein the predetermined loss function is defined according to the distance between the reconstruction data and the input data.
In one embodiment, the training module 230, when updating the parameters of the self-encoder with a predetermined loss function, may:
the parameters of the self-encoder are updated in the direction of the decrease of the predetermined loss function by the gradient descent method.
The multi-type energy demand prediction model training device can implement the multi-type energy demand prediction model training method. The specific limitation and the rest of the embodiments of the device for training the multi-type energy demand prediction model can be found in the content of the method for training the multi-type energy demand prediction model, and the embodiments are not described in detail.
Example III
Referring to fig. 4, fig. 4 is a flow chart of a method for predicting multi-type energy demand according to an embodiment of the present invention.
The multi-type energy demand prediction method in the embodiment of the invention can be applied to computer equipment such as a server.
In one embodiment, the present invention provides a method for predicting multi-type energy demand, comprising the steps of:
step S310, obtaining the energy demand influence data to be predicted.
In one embodiment, the obtained energy demand impact data to be predicted may include data such as meteorological data, time data, power energy data, grid scheduling data, socioeconomic data, carbon emission data, etc.; optionally, the data of meteorological data, time data, electric power energy data, power grid dispatching data, socioeconomic data, carbon emission data and the like are data corresponding to the current time when the to-be-predicted energy demand influence data is acquired.
In one embodiment, the computer device may obtain the energy requirement impact data to be predicted through the input of a worker.
Step S320, the trained encoder of the self-encoder is utilized to perform feature extraction on the energy demand influence data to be predicted, so as to obtain target energy demand influence data.
It will be appreciated that the trained self-encoder is the trained self-encoder in the first embodiment.
Step S330, inputting the target energy demand influence data into the multi-type energy demand prediction model to predict the target multi-type energy demand data.
It can be understood that the multi-type energy demand prediction model is the multi-type energy demand prediction model obtained through training in the first embodiment.
In one embodiment, the target multi-type energy demand data may include energy demand data such as power demand data, coal demand data, oil demand data, natural gas demand data, and the like.
For the multi-type energy demand prediction model, refer to the content in the first embodiment, and the description of the embodiment is omitted.
According to the multi-type energy demand prediction method, the obtained energy demand influence data to be predicted is subjected to feature extraction through the trained encoder of the self-encoder, the effective features of the data are extracted, and the target energy demand influence data are obtained; the target energy demand influence data is input into the multi-type energy demand prediction model to predict and obtain target multi-type energy demand data, and the multi-type energy demand prediction model has better generalization capability and more accurate energy demand data prediction capability, so that the finally predicted energy demand data is more accurate, the model has the capability of predicting and obtaining multi-type energy demand data, and the multi-type energy demand data can be finally predicted and obtained, so that the effect of energy demand prediction is greatly improved.
Example IV
In order to execute the method corresponding to the third embodiment to achieve the corresponding functions and technical effects, a multi-type energy demand prediction device is provided below.
Referring to fig. 5, fig. 5 is a block diagram illustrating a structure of a multi-type energy demand prediction apparatus according to an embodiment of the present invention.
In one embodiment, the multi-type energy demand prediction apparatus of the present invention includes:
an acquisition module 410, configured to acquire energy demand impact data to be predicted;
the feature extraction module 420 is configured to perform feature extraction on the energy demand influence data to be predicted by using the trained encoder of the self-encoder, so as to obtain target energy demand influence data;
the energy demand prediction module 430 is configured to input the target energy demand influence data into the multi-type energy demand prediction model, and predict the target multi-type energy demand data.
According to the multi-type energy demand prediction device, the obtained energy demand influence data to be predicted is subjected to feature extraction through the trained encoder of the self-encoder, the effective features of the data are extracted, and the target energy demand influence data are obtained; the target energy demand influence data is input into the multi-type energy demand prediction model to predict and obtain target multi-type energy demand data, and the multi-type energy demand prediction model has better generalization capability and more accurate energy demand data prediction capability, so that the finally predicted energy demand data is more accurate, the model has the capability of predicting and obtaining multi-type energy demand data, and the multi-type energy demand data can be finally predicted and obtained, so that the effect of energy demand prediction is greatly improved.
The above-described multi-type energy demand prediction apparatus may implement the multi-type energy demand prediction method described above. The specific limitation and the rest of the embodiments of the above-mentioned multi-type energy demand prediction device can be referred to the content of the multi-type energy demand prediction method, and the embodiments are not repeated.
Example five
In one embodiment, the present invention provides a computer device, including a memory and a processor, where the memory is configured to store a computer program, and the processor is configured to execute the computer program to cause the computer device to perform the above-described multi-type energy demand prediction model training method, or the above-described multi-type energy demand prediction method.
Alternatively, the computer device may be a server.
In one embodiment, the internal structure of the computer device of the present invention may be as shown in fig. 6.
In one embodiment, the present invention provides a computer readable storage medium storing a computer program which when executed by a processor implements the multi-type energy demand prediction model training method described above, or the multi-type energy demand prediction method described above.
In the several embodiments provided in the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. The apparatus embodiments described above are merely illustrative, for example, of the flowcharts and block diagrams in the figures that illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, functional modules in the embodiments of the present invention may be integrated together to form a single part, or each module may exist alone, or two or more modules may be integrated to form a single part.
The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, randomAccess Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The above description is only an example of the present invention and is not intended to limit the scope of the present invention, and various modifications and variations will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention. It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures.
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
Claims (10)
1. A training method for a multi-type energy demand prediction model is characterized by comprising the following steps:
acquiring a first energy demand influence data set and a corresponding multi-type energy demand data set;
performing feature extraction on the first energy demand influence data set by using a trained encoder of the self-encoder to obtain a second energy demand influence data set;
and training the time sequence prediction model to be trained by using the second energy demand influence data set and the multi-type energy demand data set to obtain the multi-type energy demand prediction model.
2. The method of claim 1, wherein after the acquiring the first energy demand impact dataset and the corresponding multi-type energy demand dataset, the feature extraction is performed on the first energy demand impact dataset using the trained encoder from the encoder, and before the obtaining the second energy demand impact dataset, the method further comprises:
training a self-encoder to be trained using the first energy demand impact dataset, the self-encoder comprising an encoder and a decoder;
-disassembling the decoder in the self-encoder.
3. The method of claim 2, wherein training the self-encoder to be trained using the first energy demand impact data set comprises:
respectively inputting the data in the first energy requirement influence data set into a self-encoder to be trained for iterative training;
in each of the iterative training steps,
the encoder of the self-encoder performs feature extraction on input data to obtain feature vectors;
the decoder of the self-encoder reconstructs the feature vector to obtain reconstructed data;
and updating the parameters of the self-encoder by using a predetermined loss function, wherein the predetermined loss function is defined according to the distance between the reconstruction data and the input data.
4. A multi-type energy demand prediction model training method as claimed in claim 3, wherein said updating parameters of said self-encoder with a predetermined loss function comprises:
and updating the parameters of the self-encoder in the direction of the reduction of the preset loss function by using a gradient descent method.
5. The method of claim 1, wherein the data in the first set of energy demand impact data comprises meteorological data, time data, electrical energy data, grid scheduling data, socioeconomic data, carbon emission data.
6. A method for predicting multi-type energy demand, comprising:
acquiring energy demand influence data to be predicted;
performing feature extraction on the energy demand influence data to be predicted by using a trained encoder of the self-encoder to obtain target energy demand influence data;
and inputting the target energy demand influence data into a multi-type energy demand prediction model, and predicting to obtain target multi-type energy demand data.
7. A multi-type energy demand prediction model training device, comprising:
the acquisition module is used for acquiring the first energy demand influence data set and the corresponding multi-type energy demand data set;
the feature extraction module is used for carrying out feature extraction on the first energy demand influence data set by utilizing the trained encoder of the self-encoder to obtain a second energy demand influence data set;
and the training module is used for training the time sequence prediction model to be trained by utilizing the second energy demand influence data set and the multi-type energy demand data set to obtain the multi-type energy demand prediction model.
8. A multi-type energy demand prediction apparatus, comprising:
the acquisition module is used for acquiring the energy demand influence data to be predicted;
the feature extraction module is used for extracting features of the energy demand influence data to be predicted by utilizing the trained encoder of the self-encoder to obtain target energy demand influence data;
and the energy demand prediction module is used for inputting the target energy demand influence data into a multi-type energy demand prediction model to predict and obtain target multi-type energy demand data.
9. A computer device comprising a memory for storing a computer program and a processor that runs the computer program to cause the computer device to perform the multi-type energy demand prediction model training method according to any one of claims 1 to 5 or the multi-type energy demand prediction method according to claim 6.
10. A computer readable storage medium, characterized in that it stores a computer program which, when executed by a processor, implements the multi-type energy demand prediction model training method according to any one of claims 1 to 5, or the multi-type energy demand prediction method according to claim 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211534854.9A CN116108960A (en) | 2022-12-01 | 2022-12-01 | Training method and device for multi-type energy demand prediction model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211534854.9A CN116108960A (en) | 2022-12-01 | 2022-12-01 | Training method and device for multi-type energy demand prediction model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116108960A true CN116108960A (en) | 2023-05-12 |
Family
ID=86256981
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211534854.9A Pending CN116108960A (en) | 2022-12-01 | 2022-12-01 | Training method and device for multi-type energy demand prediction model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116108960A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117077871A (en) * | 2023-10-17 | 2023-11-17 | 山东理工昊明新能源有限公司 | Method and device for constructing energy demand prediction model based on big data |
-
2022
- 2022-12-01 CN CN202211534854.9A patent/CN116108960A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117077871A (en) * | 2023-10-17 | 2023-11-17 | 山东理工昊明新能源有限公司 | Method and device for constructing energy demand prediction model based on big data |
CN117077871B (en) * | 2023-10-17 | 2024-02-02 | 山东理工昊明新能源有限公司 | Method and device for constructing energy demand prediction model based on big data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111091233B (en) | Short-term wind power prediction modeling method for wind power plant | |
Alencar et al. | Hybrid approach combining SARIMA and neural networks for multi-step ahead wind speed forecasting in Brazil | |
CN111080032A (en) | Load prediction method based on Transformer structure | |
Xiong et al. | A short-term wind power forecast method via xgboost hyper-parameters optimization | |
CN111242351A (en) | Tropical cyclone track prediction method based on self-encoder and GRU neural network | |
CN115758255B (en) | Power consumption abnormal behavior analysis method and device under fusion model | |
CN116108960A (en) | Training method and device for multi-type energy demand prediction model | |
CN113962433A (en) | Wind power prediction method and system fusing causal convolution and separable time convolution | |
CN110766215B (en) | Wind power climbing event prediction method based on feature adaptive selection and WDNN | |
CN112712855B (en) | Joint training-based clustering method for gene microarray containing deletion value | |
CN111193254B (en) | Residential daily electricity load prediction method and device | |
CN116596169A (en) | Power system prediction method, device and storage medium | |
CN117154680A (en) | Wind power prediction method based on non-stationary transducer model | |
CN115296298A (en) | Wind power plant power prediction method | |
CN113159437B (en) | Method for predicting short-term photovoltaic power generation output power | |
CN105469601A (en) | A road traffic space data compression method based on LZW coding | |
CN113780644A (en) | Photovoltaic output prediction method based on online learning | |
CN115860281B (en) | Multi-entity load prediction method and device for energy system based on cross-entity attention | |
CN117591942B (en) | Power load data anomaly detection method, system, medium and equipment | |
CN115310657A (en) | Online distributed photovoltaic power generation prediction method based on lstm neural network | |
Zhang et al. | Small Sample Data Augmentation Method for Photovoltaic Power Generation Based on Improved Variational Auto-encoder | |
CN117725481A (en) | Wind power probability interval prediction method, device and medium | |
CN118134021A (en) | Wind power prediction method and system in severe weather | |
CN117691565A (en) | Method, device, equipment and medium for predicting generated energy of wind power station | |
Liao et al. | Application of a novel generative adversarial network to wind power forecasting |
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 |