CN114781759A - Resident load prediction method and device based on neural network and dynamic mirror image reduction - Google Patents
Resident load prediction method and device based on neural network and dynamic mirror image reduction Download PDFInfo
- Publication number
- CN114781759A CN114781759A CN202210678689.8A CN202210678689A CN114781759A CN 114781759 A CN114781759 A CN 114781759A CN 202210678689 A CN202210678689 A CN 202210678689A CN 114781759 A CN114781759 A CN 114781759A
- Authority
- CN
- China
- Prior art keywords
- neural network
- resident
- short term
- term memory
- network model
- 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 46
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 22
- 230000009467 reduction Effects 0.000 title claims abstract description 11
- 230000005611 electricity Effects 0.000 claims abstract description 46
- 238000003062 neural network model Methods 0.000 claims abstract description 39
- 238000012549 training Methods 0.000 claims abstract description 29
- 230000006403 short-term memory Effects 0.000 claims abstract description 26
- 230000007787 long-term memory Effects 0.000 claims abstract description 25
- 230000015654 memory Effects 0.000 claims abstract description 19
- 238000012360 testing method Methods 0.000 claims abstract description 16
- 230000001105 regulatory effect Effects 0.000 claims abstract description 3
- 230000006870 function Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 12
- 210000002569 neuron Anatomy 0.000 claims description 7
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 4
- 238000012886 linear function Methods 0.000 claims description 4
- 238000011425 standardization method Methods 0.000 claims description 4
- 230000015556 catabolic process Effects 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 3
- 238000006731 degradation reaction Methods 0.000 claims description 3
- 230000007423 decrease Effects 0.000 claims 1
- 238000007430 reference method Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 230000006399 behavior Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Marketing (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a resident load prediction method based on neural network and dynamic mirror image reduction, which comprises the following steps: acquiring resident load related data sets at a plurality of moments before the moment to be predicted, carrying out regularization on the resident load related data sets, and segmenting the regularized resident load related data sets to obtain a test set and a training set; constructing a deep long and short term memory neural network model, and training the deep long and short term memory neural network model by using a training set to obtain a trained deep long and short term memory neural network model; obtaining an error adjusting variable by using an improved dynamic mirror image descent method so as to correct a predicted error; deep long-short term memory neural network using test set and trainingModel to obtain future of resident usertThe intermediate predicted value of the electricity consumption at any moment is combined with the updated value of the error regulating variable to obtain the future of the resident usertThe final predicted value of the power consumption at the moment; the method can obviously improve the load prediction precision of single residents in a short period.
Description
Technical Field
The invention belongs to the technical field of load prediction, and particularly relates to a residential load prediction method and device based on neural network and dynamic mirror image reduction.
Background
With the wide deployment of high-grade measuring facilities in modern power systems, especially smart meters, more and more residential electricity consumption data are acquired on a large scale. Such a huge amount of data enables grid companies to encourage residential users to actively participate in demand-side management, such as time-of-use electricity prices, through various demand response programs. As an important component of demand side management, residential load prediction is a very important and challenging task for power grid companies because residential loads have strong irregularities and uncertainties. Therefore, solving the residential load prediction problem plays a crucial role in the interaction between the power grid company and the residential users, efficient and economic power grid operation, and household energy consumption optimization.
Residential load predictions generally fall into two categories: user cluster prediction and single user prediction. At present, the load prediction of a residential user cluster obtains a relatively ideal prediction precision level, and the main reason is that various electricity utilization behaviors of a large number of residential users can smooth the total load curve of the residential users, so that an easily-recognized electricity utilization mode is generated; however, the accuracy level of single resident load prediction is still low compared to the resident user cluster load prediction, and needs to be further improved. In recent years, deep neural networks have been widely used for single resident user load prediction due to their superior ability to extract complex patterns. Although most of the existing deep neural network models have higher prediction accuracy than many traditional machine learning models, the existing deep neural network models are basically trained offline on limited population load data and then applied to online prediction. Therefore, these well-trained offline models may encounter many sudden load changes that are not included in the training when performing online prediction. This is mainly because the load of a single residential user can be extremely unstable and uncertain, which can have a significant negative impact on the prediction performance of the deep neural network model.
Disclosure of Invention
The technical problem to be solved by the invention is realized by adopting the following technical scheme:
the resident load prediction method based on the neural network and the dynamic mirror image reduction comprises the following steps:
acquiring resident load related data sets at a plurality of moments before the moment to be predicted, carrying out regularization on the resident load related data sets, and segmenting the regularized resident load related data sets to obtain a test set and a training set;
constructing a deep long and short term memory neural network model, and training the deep long and short term memory neural network model by using a training set to obtain a trained deep long and short term memory neural network model;
obtaining an error adjusting variable by using an improved dynamic mirror image descent method so as to correct a predicted error;
obtaining future of resident user by using test set and trained deep long-short term memory neural network modeltIntermediate predicted value of power consumption at any moment
Combined with updated values of error-adjusting variables
To obtain the future of the resident usertFinal predicted value of power consumption at time
：
Further, the deep long-short term memory neural network model comprises 3 layers of LSTM layers and 1 layer of full-connection layers, the number of neurons of the LSTM layers and the number of neurons of the full-connection layers are set to be 64, the activation functions of the LSTM layers and the full-connection layers are respectively a tanh function and a linear function, the network parameter optimizer is Adam, the loss function adopts root mean square error, and the batch size is set to be 128.
Further, an improved dynamic mirror image descent method is utilized to obtain an error adjustment variable
Thus, the method of correcting the prediction error is:
wherein, intIn the case of time point =1,k 1initialization is 0;
is a constant value of the step size,
and
respectively as residential userst-1 actual and final predicted values of electricity usage at time.
Further, the resident load related data includes electricity consumption of the resident user, electricity consumption week type, electricity consumption holiday type, and electricity consumption time information.
Furthermore, regularization processing is carried out on the electricity consumption data of the residential users by a min-max standardization method, and regularization processing is carried out on the electricity consumption week type, the electricity consumption holiday type and the electricity consumption time information of the residential users by a one-hot coding method.
The resident load prediction device based on neural network and dynamic mirror image descent comprises:
the data set module acquisition module is used for acquiring resident load related data sets at a plurality of moments before the moment to be predicted, carrying out regularization processing on the resident load related data sets, and segmenting the normalized resident load related data sets to obtain a test set and a training set;
the deep long and short term memory neural network model building and training module is used for building a deep long and short term memory neural network model and training the deep long and short term memory neural network model by using a training set to obtain a trained deep long and short term memory neural network model;
the prediction error acquisition module is used for acquiring an error adjusting variable by utilizing an improved dynamic mirror image reduction method so as to correct a prediction error;
the resident electricity consumption prediction module is used for obtaining the future of the resident user by utilizing the test set and the trained deep long-short term memory neural network modeltIntermediate predicted value of power consumption at any moment
In combination with updated values of the error regulating variable
To obtain the future of the resident usertFinal predicted value of power consumption at any moment
：
An electronic device, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the neural network and dynamic image reduction based residential load prediction method as described above.
A computer-readable storage medium storing a computer program which, when executed by a processor, implements the neural network and dynamic mirror reduction-based residential load prediction method as described above.
The invention has the advantages and positive effects that:
aiming at the problem of changeability of single resident load, the invention provides a single resident load self-adaptive prediction method based on a deep Long Short Term Memory (LSTM) neural network and a Dynamic mirror image reduction (DMD). firstly, a comprehensive characteristic expression strategy is designed to describe the load characteristic of each moment in detail, thereby forming the input of a prediction model; secondly, improving an original dynamic mirror image descent algorithm to enable the algorithm to be used for adjusting single resident load prediction; finally, a deep long-term and short-term memory neural network and a dynamic mirror image descent algorithm are fused to realize the self-adaptive prediction of the load of a single resident; the method can remarkably improve the single resident load prediction precision in a short period, and has great practical application value and popularization prospect on demand side management.
Drawings
The technical solutions of the present invention will be described in further detail below with reference to the accompanying drawings and examples, but it should be understood that these drawings are designed for illustrative purposes only and thus do not limit the scope of the present invention. Furthermore, unless otherwise indicated, the drawings are intended to be illustrative of the structural configurations described herein only, and are not necessarily drawn to scale.
FIG. 1 is a percentage improvement statistic of a solution provided by an embodiment of the present invention compared to a baseline method;
fig. 2 is a predicted load curve of the solution and the reference method according to the embodiment of the present invention.
Detailed Description
First, it should be noted that the specific structures, features, advantages, etc. of the present invention will be specifically described below by way of example, but all the descriptions are for illustrative purposes only and should not be construed as limiting the present invention in any way. Furthermore, any single feature described or implicit in any embodiment or any single feature shown or implicit in any drawing may still be combined or subtracted between any of the features (or equivalents thereof) to obtain still further embodiments of the invention that may not be directly mentioned herein.
It should be noted that, in the present application, the embodiments and features of the embodiments may be combined with each other without conflict.
The residential load prediction method based on the neural network and the dynamic image degradation comprises the following steps:
acquiring resident load related data sets at a plurality of moments before the moment to be predicted, carrying out regularization on the resident load related data sets, and segmenting the regularized resident load related data sets to obtain a test set and a training set;
constructing a deep long and short term memory neural network model, and training the deep long and short term memory neural network model by using a training set to obtain a trained deep long and short term memory neural network model;
obtaining an error adjusting variable by using an improved dynamic mirror image descending method so as to correct a prediction error;
obtaining future of resident user by using test set and trained deep long-short term memory neural network modeltIntermediate predicted value of power consumption at any moment
Combined with updated values of error-adjusting variables
To obtain the future of the resident usertFinal predicted value of power consumption at any moment
：
Specifically, the deep long-short term memory neural network model comprises 3 layers of LSTM layers and 1 layer of full-connection layers, the number of neurons of the LSTM layers and the number of neurons of the full-connection layers are set to be 64, the activation functions of the LSTM layers and the full-connection layers are respectively a tanh function and a linear function, the network parameter optimizer is Adam, the loss function adopts root mean square error, and the batch size is set to be 128.
In addition, error adjusting variables are obtained by using an improved dynamic mirror image descending method
Thus, the method for correcting the prediction error comprises the following steps:
wherein, intIn the case of time-point 1' = time-point,k 1initializing to 0;
in order to be the step-size constant,
and
are respectively residential userst-1 actual and final predicted values of electricity usage at a time;
the resident load related data comprises electricity consumption, electricity consumption week types, electricity consumption holiday types and electricity consumption time information of resident users; the electricity consumption week type represents the days from Monday to Sunday, the electricity consumption holiday type represents whether the acquired resident load related data is acquired from holidays, and the electricity consumption time information represents the time when the acquired resident load related data is acquired from holidays; it is considered that the min-max standardization method is used for conducting regularization processing on electricity consumption data of the residential users, and the one-hot coding method is used for conducting regularization processing on electricity utilization week types, electricity utilization holiday types and electricity utilization time information of the residential users.
For example, in this embodiment, taking an erland public data set (Smart metric electric Customer trihoviouer trills) as an example, 750 residential users with complete sampling data are selected, and the technical solution of the present invention is further described. Where the sampling frequency of the data set is once every half hour and for each residential user 90% of the samples are used for training and the remaining 10% are used for testing. The technical scheme of the invention is further described in detail by combining the attached drawings 1 and 2:
step 1: the method comprises the following steps of constructing input sample data which comprises 4 kinds of information of electricity consumption, week types, holidays and time information of residential users, and specifically representing the following steps:
1) for the pastTIndividual time, residential electricity load sequence
Expressed as:
wherein, the first and the second end of the pipe are connected with each other,
to be at the moment of timetElectricity consumption of residents;
2) for the pastTTime of day, time of day index sequence
Expressed as:
wherein, the first and the second end of the pipe are connected with each other,
is a time of daytThe index of (a) is stored in the database,Fthe sampling frequency for the electricity consumption of residents is generally 24, 48 or 96;
3) for the pastTTime of day, week type sequence
Expressed as:
wherein, the first and the second end of the pipe are connected with each other,
is a time of daytThe week type of (d);
4) for the pastTTime, holiday and festival mark sequence
Expressed as:
wherein the content of the first and second substances,h t is a time of daytThe holiday flag of (a) takes a value of 1 or 2, and 1 represents a non-holiday and 2 represents a holiday.
Thus, for a certain timetInputting a sample S t Consisting of the above 4 sequences, denoted:
wherein the content of the first and second substances,Et T、Dt T、Wt T、Ht Tare respectively Et、Dt、Wt、HtThe transposed matrix of (2).
And 2, step: in order to accelerate the training speed of the prediction model and improve the generalization capability of the model, the min-max standardization method and the one-hot coding method are utilized to carry out on the original input sample S t Regularizing to make each element in
Within the range; in particular, the min-max normalization method was applied to EtRegularization is carried out, and a one-hot coding method is applied to Dt、Wt、HtAnd carrying out regularization processing. Hence, regularized input samples
Expressed as:
wherein, the first and the second end of the pipe are connected with each other,
is composed of
A matrix of
，
、
、
、
Are each Et T、Dt T、Wt T、Ht TThe regularization matrix of (a); so, regularized input samples
Each row of (a) represents a detailed feature of the corresponding time instant.
And 3, step 3: and (3) forming a training sample set according to the steps 1 and 2, and constructing and training a deep long-short term memory neural network model for short-term single resident load prediction. The neural network model comprises 3 LSTM layers and 1 fully-connected layer, the number of neurons of the LSTM layers and the number of neurons of the fully-connected layer are set to be 64, a tanh function and a linear function are selected to be respectively used as activation functions of the LSTM layers and the fully-connected layer, Adam is selected to be used as a network parameter optimizer, the learning rate is set to be 0.001, Root Mean Square Error (RMSE) is selected to be used as a loss function, the batch size is set to be 128, and the input duration is set to be 24.
And 4, step 4: generating a regularized input sample according to formulas (1) to (6) by using the obtained test set in the online electricity utilization information of the single residential user, and inputting the trained deep long-short term memory neural network model to obtain the future time of the residential usertIntermediate predicted value of electricity consumption
；
And 5: the dynamic mirror down method is improved by equation (7) to update the error adjustment variable (wheretIn the case of time point =1,k 1initialization to 0), correcting the prediction error, which is expressed as follows:
in the formula (I), the compound is shown in the specification,
being a step constant, can generally beThe value is 1.0 × 10-5、1.0×10-4、1.0×10-3、1.0×10-2、1.0×10-1Or 1.0X 100，
And
are respectively residential userst-1 actual and final predicted values of electricity usage at a time;
step 6: according to a formula (8), fusing the intermediate predicted value of the electricity consumption of the residential user at the future time
And updated values of error adjusting variables
Calculating the final predicted value of the electricity consumption of the residential user at the future time
The specific expression is as follows:
wherein the content of the first and second substances,
is a resident usertThe final predicted value of the power consumption at the moment.
Table 1 shows the results of the residential load prediction of the reference method and the solution of the present invention, fig. 1 shows the statistics of the percentage increase of the solution of the present invention compared to the reference method, fig. 2 shows the predicted load curve of the solution of the present invention and the reference method, and it can be seen from both fig. 1 and 2 that the solution of the present invention has better effect than other methods; as can be seen from table 1, the mean absolute error and the root mean square error of the method of the present invention are improved compared to the baseline method.
TABLE 1 resident load prediction results of the benchmark method and the inventive scheme
Example 2
Based on the same inventive concept, the embodiment of the application also provides a resident load prediction device based on neural network and dynamic mirror descent, which comprises:
the data set module acquisition module is used for acquiring resident load related data sets at a plurality of moments before the moment to be predicted, carrying out regularization processing on the resident load related data sets, and segmenting the normalized resident load related data sets to obtain a test set and a training set;
the deep long and short term memory neural network model building and training module is used for building a deep long and short term memory neural network model and training the deep long and short term memory neural network model by using a training set to obtain a trained deep long and short term memory neural network model;
the prediction error acquisition module is used for acquiring an error adjusting variable by utilizing an improved dynamic mirror image descent method so as to correct a prediction error;
the resident electricity consumption prediction module is used for obtaining the future of the resident user by utilizing the test set and the trained deep long-short term memory neural network modeltIntermediate predicted value of power consumption at any moment
Combined with updated values of error-adjusting variables
To obtain the future of the resident usertFinal predicted value of power consumption at time
：
Based on the same inventive concept, an embodiment of the present application further provides an electronic device, including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the neural network and dynamic image descent-based residential load prediction method described above; it should be noted that the electronic device may include, but is not limited to, a processing unit and a storage unit; those skilled in the art will appreciate that the electronic device including the processing unit, the memory unit do not constitute a limitation of the computing device, may include more components, or combine certain components, or different components, for example, the electronic device may also include an input output device, a network access device, a bus, etc.
A computer-readable storage medium storing a computer program which, when executed by a processor, implements the above-described residential load prediction method based on neural network and dynamic mirror down; it should be noted that the readable storage medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof; the program embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing. For example, program code for carrying out operations of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the C programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, or entirely on a remote computing device or server. In situations involving remote computing devices, the remote computing devices may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to external computing devices (e.g., through the internet using an internet service provider).
The present invention has been described in detail with reference to the above examples, but the description is only for the preferred examples of the present invention and should not be construed as limiting the scope of the present invention. All equivalent changes and modifications made within the scope of the present invention shall fall within the scope of the present invention.
Claims (8)
1. The resident load prediction method based on the neural network and the dynamic mirror image reduction is characterized by comprising the following steps of:
acquiring resident load related data sets at a plurality of moments before the moment to be predicted, carrying out regularization on the resident load related data sets, and segmenting the regularized resident load related data sets to obtain a test set and a training set;
constructing a deep long and short term memory neural network model, and training the deep long and short term memory neural network model by using a training set to obtain a trained deep long and short term memory neural network model;
obtaining an error adjusting variable by using an improved dynamic mirror image descent method so as to correct a predicted error;
obtaining future of resident user by using test set and trained deep long-short term memory neural network modeltIntermediate predicted value of power consumption at any moment
Combined with updated values of error-adjusting variables
To obtain the future of the resident usertFinal predicted value of power consumption at any moment
：
2. The residential load prediction method based on neural network and dynamic mirror descent as claimed in claim 1, wherein: the deep long-short term memory neural network model comprises 3 LSTM layers and 1 full-connection layer, the number of neurons of the LSTM layers and the full-connection layers is set to be 64, the activation functions of the LSTM layers and the full-connection layers are respectively a tanh function and a linear function, a network parameter optimizer is Adam, the loss function adopts root-mean-square error, and the batch size is set to be 128.
3. The residential load prediction method based on neural network and dynamic mirror descent as claimed in claim 1, wherein: obtaining error adjustment variables using an improved dynamic mirror descent method
Thus, the method of correcting the prediction error is:
wherein, intIn the case of time point =1,k 1initialization is 0;
in order to be the step-size constant,
and
respectively as residential userst-1 actual and final predicted values of electricity usage at time.
4. The residential load prediction method based on neural network and dynamic image degradation according to claim 1, characterized in that: the resident load related data comprises electricity consumption, electricity consumption week type, electricity consumption holiday type and electricity consumption time information of the resident users.
5. The residential load prediction method based on neural network and dynamic image degradation according to claim 4, characterized in that: the method comprises the steps of utilizing a min-max standardization method to conduct regularization processing on electricity consumption data of residential users, and utilizing a one-hot coding method to conduct regularization processing on electricity utilization week types, electricity utilization holiday types and electricity utilization time information of the residential users.
6. Resident load prediction device based on neural network and dynamic mirror image decline characterized by includes:
the data set module acquisition module is used for acquiring resident load related data sets at a plurality of moments before the moment to be predicted, carrying out regularization processing on the resident load related data sets, and segmenting the normalized resident load related data sets to obtain a test set and a training set;
the deep long and short term memory neural network model building and training module is used for building a deep long and short term memory neural network model and training the deep long and short term memory neural network model by using a training set to obtain a trained deep long and short term memory neural network model;
the prediction error acquisition module is used for acquiring an error adjusting variable by utilizing an improved dynamic mirror image reduction method so as to correct a prediction error;
a prediction module for residential electricity consumption for obtaining the future of residential users by using the test set and the trained deep long-short term memory neural network modeltIntermediate predicted value of power consumption at any moment
In combination with updated values of the error regulating variable
To obtain the future of the resident usertElectricity consumption at any momentFinal predicted value of quantity
：
7. An electronic device, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 5.
8. A computer-readable storage medium, storing a computer program which, when executed by a processor, implements the method of any of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210678689.8A CN114781759A (en) | 2022-06-16 | 2022-06-16 | Resident load prediction method and device based on neural network and dynamic mirror image reduction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210678689.8A CN114781759A (en) | 2022-06-16 | 2022-06-16 | Resident load prediction method and device based on neural network and dynamic mirror image reduction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114781759A true CN114781759A (en) | 2022-07-22 |
Family
ID=82421315
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210678689.8A Pending CN114781759A (en) | 2022-06-16 | 2022-06-16 | Resident load prediction method and device based on neural network and dynamic mirror image reduction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114781759A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109100429A (en) * | 2018-06-26 | 2018-12-28 | 武汉大学 | A kind of oil dissolved gas prediction technique of combined prediction residual GM |
| CN109754113A (en) * | 2018-11-29 | 2019-05-14 | 南京邮电大学 | Load forecasting method based on dynamic time warping Yu length time memory |
| CN109800898A (en) * | 2017-11-17 | 2019-05-24 | 中国电力科学研究院有限公司 | A kind of intelligence short-term load forecasting method and system |
| CN111489036A (en) * | 2020-04-14 | 2020-08-04 | 天津相和电气科技有限公司 | Resident load prediction method and device based on electrical appliance load characteristics and deep learning |
| CN112446537A (en) * | 2020-11-20 | 2021-03-05 | 国网浙江省电力有限公司宁波供电公司 | Short-term load prediction method based on deep long-term and short-term memory network |
-
2022
- 2022-06-16 CN CN202210678689.8A patent/CN114781759A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109800898A (en) * | 2017-11-17 | 2019-05-24 | 中国电力科学研究院有限公司 | A kind of intelligence short-term load forecasting method and system |
| CN109100429A (en) * | 2018-06-26 | 2018-12-28 | 武汉大学 | A kind of oil dissolved gas prediction technique of combined prediction residual GM |
| CN109754113A (en) * | 2018-11-29 | 2019-05-14 | 南京邮电大学 | Load forecasting method based on dynamic time warping Yu length time memory |
| CN111489036A (en) * | 2020-04-14 | 2020-08-04 | 天津相和电气科技有限公司 | Resident load prediction method and device based on electrical appliance load characteristics and deep learning |
| CN112446537A (en) * | 2020-11-20 | 2021-03-05 | 国网浙江省电力有限公司宁波供电公司 | Short-term load prediction method based on deep long-term and short-term memory network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | A seasonal GM (1, 1) model for forecasting the electricity consumption of the primary economic sectors | |
| CN110969285B (en) | Prediction model training method, prediction device, prediction equipment and medium | |
| Lou et al. | A novel random fuzzy neural networks for tackling uncertainties of electric load forecasting | |
| CN110610280A (en) | Short-term prediction method, model, device and system for power load | |
| Lei et al. | A proposed grey model for short-term electricity price forecasting in competitive power markets | |
| US20190312457A1 (en) | System and method for dynamic energy storage system control | |
| CN111210093B (en) | Daily water consumption prediction method based on big data | |
| CN115169703A (en) | Short-term power load prediction method based on long-term and short-term memory network combination | |
| Aouad et al. | A CNN-Sequence-to-Sequence network with attention for residential short-term load forecasting | |
| CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
| CN114330934A (en) | Model parameter self-adaptive GRU new energy short-term power generation power prediction method | |
| Bhotto et al. | Short-term demand prediction using an ensemble of linearly-constrained estimators | |
| CN114692981A (en) | Medium-and-long-term runoff forecasting method and system based on Seq2Seq model | |
| Yu et al. | Short-term cooling and heating loads forecasting of building district energy system based on data-driven models | |
| CN110705738B (en) | Intelligent electricity utilization stimulation demand response method and system based on artificial intelligence | |
| CN116011633B (en) | Regional gas consumption prediction method, regional gas consumption prediction system, regional gas consumption prediction equipment and Internet of things cloud platform | |
| CN112101612A (en) | Prediction method of town gas usage, storage medium and terminal equipment | |
| CN114781759A (en) | Resident load prediction method and device based on neural network and dynamic mirror image reduction | |
| CN110659775A (en) | LSTM-based improved electric power short-time load prediction algorithm | |
| CN114862032B (en) | XGBoost-LSTM-based power grid load prediction method and device | |
| Wang et al. | Adaptive forecasting of diverse electrical and heating loads in community integrated energy system based on deep transfer learning | |
| Irankhah et al. | A parallel CNN-BiGRU network for short-term load forecasting in demand-side management | |
| Wang et al. | An electricity load forecasting model based on multilayer dilated LSTM network and attention mechanism | |
| Han et al. | Adaptive individual residential load forecasting based on deep learning and dynamic mirror descent | |
| Li et al. | Decomposition-accumulation principle-based monthly electricity consumption forecasting approach using emd-xgboost hybrid model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220722 |
|
| RJ01 | Rejection of invention patent application after publication |