CN117744855A - Load prediction system and method based on machine learning - Google Patents
Load prediction system and method based on machine learning Download PDFInfo
- Publication number
- CN117744855A CN117744855A CN202311567701.9A CN202311567701A CN117744855A CN 117744855 A CN117744855 A CN 117744855A CN 202311567701 A CN202311567701 A CN 202311567701A CN 117744855 A CN117744855 A CN 117744855A
- Authority
- CN
- China
- Prior art keywords
- data
- historical
- feature
- vector
- load
- 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
- 238000010801 machine learning Methods 0.000 title claims abstract description 69
- 238000000034 method Methods 0.000 title claims abstract description 25
- 239000013598 vector Substances 0.000 claims description 237
- 230000005611 electricity Effects 0.000 claims description 165
- 239000011159 matrix material Substances 0.000 claims description 70
- 238000012545 processing Methods 0.000 claims description 28
- 238000013527 convolutional neural network Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 19
- 238000003062 neural network model Methods 0.000 claims description 17
- 238000010586 diagram Methods 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 14
- 230000004927 fusion Effects 0.000 claims description 14
- 230000004913 activation Effects 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 11
- 238000011176 pooling Methods 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 230000003213 activating effect Effects 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 238000004590 computer program Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000010248 power generation Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
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
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the field of intelligent prediction, and particularly discloses a load prediction system and method based on machine learning.
Description
Technical Field
The present application relates to the field of intelligent prediction, and more particularly, to a machine learning-based load prediction system and method.
Background
Load prediction is an important task in electrical power systems that aims to accurately estimate the power demand over a certain period of time in the future. This predictive effort is critical to utility companies and energy suppliers because they need to rationally arrange power generation plans, resource allocation, and grid operation to ensure that the customer's needs can be met and that the power system remains stable.
With the development of society and the increasing demand of electricity by people, load prediction is becoming more and more important. By accurately predicting the future load demand, the utility company can reasonably plan the input and output of the power generation equipment to ensure that the power demand of the user can be met in the peak period. At the same time, this also helps to avoid an insufficient or excessive supply of electric power, thereby improving energy utilization efficiency and reducing costs.
Therefore, a load prediction scheme based on machine learning is required.
Disclosure of Invention
The present application has been made in order to solve the above technical problems. The embodiment of the application provides a load prediction system and a load prediction method based on machine learning, which are characterized in that the relevant characteristics of historical power consumption load are obtained by carrying out characteristic extraction on historical weather data, temperature data, power consumption data and date data, the relevant data characteristics affecting the power consumption are obtained by carrying out characteristic extraction on the weather data, the temperature data and the date data of the same day, and the extracted characteristics are associated in a high-dimensional space and then are decoded by a decoder to obtain a decoding value for representing a prediction result of the power consumption of the same day.
According to one aspect of the present application, there is provided a machine learning based load prediction system comprising:
The data acquisition module is used for acquiring historical weather data, temperature data, date data, power consumption data, and weather data, temperature data and date data of the same day;
the electricity consumption prediction feature extraction module is used for arranging the weather data, the temperature data and the date data of the current day into an electricity consumption influence parameter matrix and then obtaining electricity consumption prediction feature vectors through a first convolution neural network model serving as a feature extractor;
the historical electricity load data coding module is used for enabling the historical weather data, temperature data, date data and electricity consumption data to pass through a context coder comprising an embedded layer to obtain a plurality of historical electricity load related data feature vectors;
the first scale historical electricity load characteristic generation module is used for cascading the plurality of historical electricity load related data characteristic vectors to obtain a first scale historical electricity load related data characteristic vector;
the second scale historical electricity load characteristic generation module is used for two-dimensionally arranging the plurality of historical electricity load related data characteristic vectors into a historical electricity load related data semantic characteristic matrix and then obtaining a second scale historical electricity load related data characteristic vector through a second convolution neural network model comprising a plurality of mixed convolution layers;
The multi-scale historical power consumption load characteristic fusion module is used for fusing the first-scale historical power consumption load related data characteristic vector and the second-scale historical power consumption load related data characteristic vector to obtain a historical power consumption load data characteristic vector;
the decoding characteristic generation module is used for carrying out order-based characteristic engineering matching on the electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector so as to obtain a decoding characteristic vector;
and the electricity consumption prediction result generation module is used for enabling the decoding feature vector to pass through a decoder to obtain a decoding value, wherein the decoding value is used for representing a predicted value of the electricity consumption of the current day.
In the above machine learning-based load prediction system, the electricity consumption prediction feature extraction module is configured to: each layer of the first convolutional neural network model serving as the feature extractor performs input data in the forward transfer process of the layer: carrying out convolution processing on the input data to obtain a convolution characteristic diagram; carrying out mean pooling treatment on each feature matrix along the channel dimension on the convolution feature map to obtain a pooled feature map; non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map; the output of the last layer of the first convolutional neural network model is the electricity consumption prediction feature vector, and the input of the first layer of the first convolutional neural network model is the electricity consumption influence parameter matrix.
In the above-mentioned load prediction system based on machine learning, the historical electricity load data encoding module includes: the word segmentation unit is used for carrying out word segmentation processing on the historical weather data, the historical temperature data, the historical date data and the historical electricity consumption data to obtain a plurality of historical electricity consumption load related words; the word embedding unit is used for enabling the historical electric load related words to pass through an embedding layer of the context encoder so as to convert each historical electric load related word in the historical electric load related words into a historical electric load related word embedding vector to obtain a sequence of the historical electric load related word embedding vector, wherein the embedding layer uses a learnable embedding matrix to carry out embedding encoding on each historical electric load related word; and the context semantic coding unit is used for inputting the sequence of the historical electricity load related word embedded vectors into a converter of the context encoder to obtain the plurality of historical electricity load related data feature vectors.
In the above-described machine learning-based load prediction system, the context semantic coding unit includes: a sequence arrangement subunit, configured to arrange the sequence of the historical electric load related word embedding vector as an input vector; the vector conversion subunit is used for respectively converting the input vector into a query vector and a key vector through a learning embedding matrix; a self-attention association matrix generation subunit, configured to calculate a product between the query vector and a transpose vector of the key vector to obtain a self-attention association matrix; the normalization processing subunit is used for performing normalization processing on the self-attention association matrix to obtain a normalized self-attention association matrix; an activating subunit, configured to activate the standardized self-attention association matrix input Softmax activating function to obtain a self-attention feature matrix; and the attention applying subunit is used for multiplying the self-attention characteristic matrix with each historical electric load related word embedded vector in the sequence of the historical electric load related word embedded vectors as a value vector to obtain the plurality of historical electric load related data characteristic vectors.
In the above machine learning-based load prediction system, the second scale historical electricity load feature generation module is configured to: each layer of the second convolutional neural network model comprising a plurality of mixed convolutional layers respectively carries out input data in the forward transmission process of the layer: performing multi-scale convolution coding on the input data to obtain a multi-scale feature map; carrying out mean pooling treatment on each feature matrix along the channel dimension on the multi-scale feature map to obtain a pooled feature map; non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map; the input of the first layer of the second convolution neural network model is the historical power consumption load related data semantic feature matrix, and the output of the last layer of the second convolution neural network model is the second scale historical power consumption load related data feature vector.
In the above machine learning based load prediction system, the second scale historical electricity load feature generation module is further configured to: performing convolution processing on the input data based on a first convolution kernel to obtain a first scale feature map; performing convolution processing on the input data based on a second convolution kernel to obtain a second scale feature map, wherein the second convolution kernel is a cavity convolution kernel with first cavity rate; performing convolution processing on the input data based on a third convolution kernel to obtain a third scale feature map, wherein the third convolution kernel is a cavity convolution kernel with a second cavity rate; performing convolution processing on the input data based on a fourth convolution kernel to obtain a fourth scale feature map, wherein the fourth convolution kernel is a cavity convolution kernel with a third cavity rate; and cascading the first scale feature map, the second scale feature map, the third scale feature map and the fourth scale feature map to obtain the multi-scale feature map.
In the above load prediction system based on machine learning, the multi-scale historical electricity load feature fusion module is configured to: fusing the first scale historical electrical load related data feature vector and the second scale historical electrical load related data feature vector using a fusion function to obtain the historical electrical load data feature vector; wherein the fusion function is expressed as:
f(X i ,X j )=Relu(W f [θ(X i ),φ(X j )])
wherein W is f ,θ(X i ) And phi (X) j ) All represent the point convolution of the input, relu as the activation function, []Representing the splicing operation, X i Characteristic values, X, representing positions in the first scale historical electrical load related data characteristic vector j Characteristic values representing respective positions in the second-scale historical electric load related data characteristic vector, f (X) i ,X j ) And representing the historical electricity load data characteristic vector.
In the above-described machine learning-based load prediction system, the decoding feature generation module includes: the characteristic engineering matching factor calculation unit is used for calculating characteristic engineering matching factors based on order between the electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector; the weighting unit is used for weighting the electricity prediction feature vector by taking the feature engineering matching factor as a weight so as to obtain a weighted electricity prediction feature vector; and the per-position weighted sum unit is used for calculating a per-position weighted sum between the weighted electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector to obtain the decoding characteristic vector.
In the above machine learning-based load prediction system, the feature engineering matching factor calculating unit is configured to: calculating a characteristic engineering matching factor based on order between the electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector according to the following calculation formula; wherein, the calculation formula is:
wherein V is 1 Representing the electricity consumption prediction characteristic vector, V 2 Representing the characteristic vector of the historical electricity load data, and T represents the transposition of the vector F The Frobenius norm of the matrix, exp (·) the exponential operation of the matrix, log the logarithmic function value based on 2, det the determinant of the matrix,alpha represents a superparameter, w 1 Representing the feature engineering matching factor.
According to another aspect of the present application, there is provided a machine learning-based load prediction method, including:
acquiring historical weather data, temperature data, date data, electricity consumption data and weather data, temperature data and date data of the same day;
the weather data, the temperature data and the date data of the current day are arranged into an electricity consumption influence parameter matrix, and then an electricity consumption prediction feature vector is obtained through a first convolution neural network model serving as a feature extractor;
Passing the historical weather data, temperature data, date data and electricity consumption data through a context encoder comprising an embedded layer to obtain a plurality of historical electricity consumption load related data feature vectors;
cascading the plurality of historical electricity load related data feature vectors to obtain a first-scale historical electricity load related data feature vector;
the historical power consumption load related data feature vectors are two-dimensionally arranged into a historical power consumption load related data semantic feature matrix, and then a second scale historical power consumption load related data feature vector is obtained through a second convolution neural network model comprising a plurality of mixed convolution layers;
fusing the first-scale historical power consumption load related data feature vector and the second-scale historical power consumption load related data feature vector to obtain a historical power consumption load data feature vector;
performing order-based feature engineering matching on the electricity consumption prediction feature vector and the historical electricity consumption load data feature vector to obtain a decoding feature vector;
and the decoding eigenvector is passed through a decoder to obtain a decoding value, wherein the decoding value is used for representing the predicted value of the current power consumption.
Compared with the prior art, the load prediction system and method based on machine learning provided by the application are characterized in that the relevant characteristics of the historical power consumption load are obtained by extracting the characteristics of the historical weather data, the temperature data, the power consumption data and the date data, the relevant data characteristics affecting the power consumption are obtained by extracting the characteristics of the weather data, the temperature data and the date data of the current day, and the extracted characteristics are associated in a high-dimensional space and then are decoded by a decoder to obtain a decoding value for representing the prediction result of the power consumption of the current day.
Drawings
The foregoing and other objects, features and advantages of the present application will become more apparent from the following more particular description of embodiments of the present application, as illustrated in the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and not to limit the application. In the drawings, like reference numerals generally refer to like parts or steps.
Fig. 1 is a system block diagram of a machine learning based load prediction system according to an embodiment of the present application.
Fig. 2 is an architecture diagram of a machine learning based load prediction system according to an embodiment of the present application.
FIG. 3 is a block diagram of a historical electrical load data encoding module in a machine learning based load prediction system according to an embodiment of the present application.
Fig. 4 is a block diagram of a context semantic coding unit in a machine learning based load prediction system according to an embodiment of the present application.
Fig. 5 is a block diagram of a decoding feature generation module in a machine learning based load prediction system according to an embodiment of the present application.
Fig. 6 is a flowchart of a machine learning based load prediction method according to an embodiment of the present application.
Fig. 7 is a block diagram of an electronic device according to an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Exemplary System
Fig. 1 is a system block diagram of a machine learning based load prediction system according to an embodiment of the present application. Fig. 2 is an architecture diagram of a machine learning based load prediction system according to an embodiment of the present application. As shown in fig. 1 and 2, in the machine learning-based load prediction system 100, there is included: a data acquisition module 110, configured to acquire historical weather data, temperature data, date data, electricity consumption data, and weather data, temperature data, and date data of the same day; the electricity consumption prediction feature extraction module 120 is configured to arrange weather data, temperature data, and date data of the current day into an electricity consumption influence parameter matrix, and then obtain an electricity consumption prediction feature vector through a first convolutional neural network model serving as a feature extractor; a historical electricity load data encoding module 130, configured to obtain a plurality of historical electricity load related data feature vectors by passing the historical weather data, temperature data, date data, and electricity consumption data through a context encoder including an embedded layer; the first scale historical electricity load feature generation module 140 is configured to concatenate the plurality of historical electricity load related data feature vectors to obtain a first scale historical electricity load related data feature vector; the second scale historical electricity load feature generation module 150 is configured to two-dimensionally arrange the plurality of historical electricity load related data feature vectors into a historical electricity load related data semantic feature matrix, and then obtain a second scale historical electricity load related data feature vector through a second convolutional neural network model including a plurality of hybrid convolutional layers; the multi-scale historical power consumption load feature fusion module 160 is configured to fuse the first-scale historical power consumption load related data feature vector and the second-scale historical power consumption load related data feature vector to obtain a historical power consumption load data feature vector; a decoding feature generation module 170, configured to perform order-based feature engineering matching on the electricity consumption prediction feature vector and the historical electricity consumption load data feature vector to obtain a decoding feature vector; and the electricity consumption prediction result generation module 180 is used for passing the decoding feature vector through a decoder to obtain a decoding value, wherein the decoding value is used for representing a predicted value of the electricity consumption of the current day.
In the machine learning based load prediction system 100, the data acquisition module 110 is configured to acquire historical weather data, temperature data, date data, and electricity consumption data, and weather data, temperature data, and date data of the current day. As discussed above in the background, load prediction is a vital task in electrical power systems that estimates future electrical power demand such that power generation plans are rationally arranged and resources are efficiently allocated. The task has important significance for keeping the stable operation of the power system and meeting the power demand of users. Meanwhile, the accurate load prediction can improve the use efficiency of energy sources, reduce the operation cost of the power system, and is beneficial to realizing the economic, efficient and safe operation of the power system. Therefore, a load prediction scheme based on machine learning is desired.
In recent years, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks have also shown levels approaching and even exceeding humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like. The development of deep learning and neural networks provides new solutions and solutions for machine learning-based load prediction.
Specifically, in the technical scheme of the application, firstly, historical weather data, temperature data, date data, electricity consumption data, and weather data, temperature data and date data of the same day are obtained. Weather data and temperature data are one of the important influencing factors in load prediction. Weather conditions and temperature variations can have an impact on the electricity usage behavior of people. For example, in hot summer, people may use electrical equipment such as air conditioners to increase power usage. Date data is also an important feature because load demand typically varies periodically over time. For example, there may be differences in electricity demand on weekends and weekdays, holidays and seasonal variations may also have an impact on the load. Historical electricity consumption data is a target variable in load prediction and is also key data for model training. By taking historical power usage data as input, the model can learn trends and patterns of power usage and make inferences based on the historical data when predicted. The weather data, temperature data and date data of the day are for real-time prediction. In practical applications, load prediction often needs to be updated in time to reflect the current day. Therefore, the latest environmental information can be provided by acquiring the weather data, the temperature data and the date data of the current day, so that the model is helped to accurately predict the load. Historical weather data and temperature data as well as weather data and temperature data of the same day can be obtained from weather offices and weather websites, historical date data and date data of the same day can be obtained from a date store, and historical electricity consumption data can be obtained from an electric company through request.
In the machine learning-based load prediction system 100, the electricity consumption prediction feature extraction module 120 is configured to arrange weather data, temperature data, and date data of the current day into an electricity consumption influence parameter matrix, and then obtain an electricity consumption prediction feature vector through a first convolutional neural network model serving as a feature extractor. It should be appreciated by those of ordinary skill in the art that convolutional neural networks perform well in feature extraction. The layering is one of the most important layers of the convolutional neural network, local features of the input data are extracted by performing convolution operation on the input data and a group of learnable convolution kernels, and the convolution operation is performed on the input data in a sliding window mode to generate a series of feature graphs; the pooling layer is used for downsampling the feature map, so that the dimension of the feature map is reduced and main features can be reserved; the activation function is an important component in convolutional neural networks, and by introducing nonlinear transformation, the expression capacity of the network and the capacity of fitting complex functions are enhanced. The weather data, temperature data and date data of the current day are arranged into a power consumption influence parameter matrix, which can be regarded as a two-dimensional image. By inputting the electricity usage influence parameter matrix into the convolutional neural network model as a feature extractor, features related to electricity usage influence can be extracted, which helps the model better understand the influence of the environmental conditions of the day on electricity usage.
Specifically, in the machine learning-based load prediction system 100, the electricity consumption prediction feature extraction module 120 is configured to: each layer of the first convolutional neural network model serving as the feature extractor performs input data in the forward transfer process of the layer: carrying out convolution processing on the input data to obtain a convolution characteristic diagram; carrying out mean pooling treatment on each feature matrix along the channel dimension on the convolution feature map to obtain a pooled feature map; non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map; the output of the last layer of the first convolutional neural network model is the electricity consumption prediction feature vector, and the input of the first layer of the first convolutional neural network model is the electricity consumption influence parameter matrix.
In the machine learning based load prediction system 100, the historical electricity load data encoding module 130 is configured to pass the historical weather data, temperature data, date data, and electricity consumption data through a context encoder including an embedded layer to obtain a plurality of historical electricity load related data feature vectors. An embedded layer is a common technique for converting discrete data into a continuous vector representation. Historical data is important for understanding and predicting future load demands. By inputting historical weather data, temperature data, date data, and electricity usage data into a context encoder that includes an embedded layer, context information in the data can be captured and dependencies between different elements in the data can be learned.
FIG. 3 is a block diagram of a historical electrical load data encoding module in a machine learning based load prediction system according to an embodiment of the present application. As shown in fig. 3, the historical electrical load data encoding module 130 includes: a word segmentation unit 131, configured to perform word segmentation processing on the historical weather data, temperature data, date data and electricity consumption data to obtain a plurality of historical electricity consumption load related words; a word embedding unit 132, configured to pass the plurality of historical electric load related words through an embedding layer of the context encoder to convert each of the plurality of historical electric load related words into a historical electric load related word embedding vector to obtain a sequence of historical electric load related word embedding vectors, where the embedding layer uses a learnable embedding matrix to perform embedded encoding on each of the historical electric load related words; a context semantic coding unit 133 for inputting the sequence of history electric load related words embedded vectors into a converter of the context encoder to obtain the plurality of history electric load related data feature vectors.
Fig. 4 is a block diagram of a context semantic coding unit in a machine learning based load prediction system according to an embodiment of the present application. As shown in fig. 4, the context semantic coding unit 133 includes: a sequence arrangement subunit 1331, configured to arrange the sequence of the historical electric load related word embedding vectors into an input vector; a vector conversion subunit 1332, configured to convert the input vector into a query vector and a key vector through a learning embedding matrix respectively; a self-attention correlation matrix generation subunit 1334 for calculating a product between the query vector and the transpose vector of the key vector to obtain a self-attention correlation matrix; a normalization processing subunit 1335, configured to perform normalization processing on the self-attention correlation matrix to obtain a normalized self-attention correlation matrix; an activation subunit 1336, configured to activate the normalized self-attention association matrix input Softmax activation function to obtain a self-attention feature matrix; an attention applying subunit 1337 configured to multiply the self-attention feature matrix with each of the history electric load related word embedding vectors in the sequence of history electric load related word embedding vectors as a value vector to obtain the plurality of history electric load related data feature vectors, respectively.
In the machine learning based load prediction system 100, the first scale historical electrical load feature generation module 140 is configured to concatenate the plurality of historical electrical load related data feature vectors to obtain a first scale historical electrical load related data feature vector. Time series information of the history data is very important. By cascading together a plurality of historical electrical load related data feature vectors, the feature information at different points in time can be combined to form a longer feature vector. This has the advantage that important features such as timing patterns, periodicity and trends in the historical data can be captured.
In the machine learning-based load prediction system 100, the second scale historical electrical load feature generating module 150 is configured to two-dimensionally arrange the plurality of historical electrical load related data feature vectors into a historical electrical load related data semantic feature matrix, and then obtain a second scale historical electrical load related data feature vector through a second convolutional neural network model including a plurality of hybrid convolutional layers. The historical electrical load related data feature vector contains historical information on different time scales but is still a one-dimensional vector representation. In order to further utilize the spatial structure in the data and the relationship between adjacent points, these feature vectors may be arranged into a two-dimensional matrix, i.e. a semantic feature matrix of historical electrical load related data. By passing the arranged historical power utilization load related data semantic feature matrix through a convolutional neural network model comprising a plurality of mixed convolutional layers, higher-level feature representations can be captured so as to better express patterns and related information in the historical data.
Specifically, in the machine learning based load prediction system 100, the second scale historical electrical load signature generation module 150 is configured to: each layer of the second convolutional neural network model comprising a plurality of mixed convolutional layers respectively carries out input data in the forward transmission process of the layer: performing multi-scale convolution coding on the input data to obtain a multi-scale feature map; carrying out mean pooling treatment on each feature matrix along the channel dimension on the multi-scale feature map to obtain a pooled feature map; non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map; the input of the first layer of the second convolution neural network model is the historical power consumption load related data semantic feature matrix, and the output of the last layer of the second convolution neural network model is the second scale historical power consumption load related data feature vector.
In the machine learning based load prediction system 100, the multi-scale historical electrical load feature fusion module 160 is configured to fuse the first-scale historical electrical load related data feature vector and the second-scale historical electrical load related data feature vector to obtain a historical electrical load data feature vector. It should be appreciated that the first scale historical electrical load related data feature vector captures historical patterns and trends over different time scales and may reflect load changes over a short period of time. The second scale historical power consumption load related data feature vector extracts a higher level feature representation through the mixed convolution layer, and can capture a historical mode and trend on a longer time scale. By fusing the historical electrical load related data feature vectors of the first scale and the second scale, information on different time scales can be mutually supplemented and cross-validated. The fused historical electricity load data feature vector comprehensively considers short-term and long-term historical modes, trends and relativity, and provides more comprehensive and richer feature representation.
Specifically, in the machine learning based load prediction system 100, the multi-scale historical electrical load feature fusion module 160 is configured to: fusing the first scale historical electrical load related data feature vector and the second scale historical electrical load related data feature vector using a fusion function to obtain the historical electrical load data feature vector; wherein the fusion function is expressed as:
f(X i ,X j )=Relu(W f [θ(X i ),φ(X j )])
wherein W is f ,θ(X i ) And phi (X) j ) All represent the point convolution of the input, relu as the activation function, []Representing the splicing operation, X i Characteristic values, X, representing positions in the first scale historical electrical load related data characteristic vector j Characteristic values representing respective positions in the second-scale historical electric load related data characteristic vector, f (X) i ,X j ) And representing the historical electricity load data characteristic vector.
In particular, in the technical scheme of the application, the electricity consumption prediction feature vector and the historical electricity consumption load data feature vector are considered to be extracted based on different data sources and processing modes. These two feature vectors are used for prediction of the amount of electricity consumption and representation of historical electricity load related data, respectively. The electricity consumption prediction feature vector is obtained by arranging weather data, temperature data and date data of the current day into an electricity consumption influence parameter matrix and processing the electricity consumption influence parameter matrix through a first convolution neural network model serving as a feature extractor. The feature vector is mainly focused on the influence of the weather, temperature and date of the day on the electricity consumption. The historical electricity load data feature vector is obtained by processing historical weather data, temperature data, date data and electricity consumption data through a context encoder comprising an embedded layer. The historical data are converted into a plurality of historical electricity load related data feature vectors through a context encoder, and the historical electricity load related data feature vectors of a first scale are obtained through cascading operation. These eigenvectors are then processed through a second convolutional neural network model comprising a plurality of mixed convolutional layers to obtain historical electrical load related data eigenvectors at a second scale. These feature vectors focus mainly on the time series features and correlations of historical electrical load data. Because the electricity prediction feature vector and the historical electricity load data feature vector are extracted based on different data processing modes, the electricity prediction feature vector and the historical electricity load data feature vector have different feature depth dimensions and filter resolutions. The electricity prediction feature vector focuses mainly on weather, temperature and date information of the current day, and the historical electricity load data feature vector focuses mainly on time series features and correlations of the historical electricity load data. This discrepancy results in the possibility of remote relationship ambiguity in the feature vector fusion. Remote relational ambiguity means that semantic associations between power consumption prediction feature vectors and historical power consumption load data feature vectors may not be accurately established, resulting in inconsistent feature distribution structures of decoded feature vectors. Such inconsistencies may affect the accuracy of the decoding decisions of the decoded feature vectors by the decoder. The decoder typically relies on the consistency and correlation of the feature vectors for decoding tasks, and if the feature distribution structure of the decoded feature vectors is not uniform, the decoder may be disturbed, resulting in a reduced accuracy of the decoding decisions.
In the machine learning based load prediction system 100, the decoding feature generation module 170 is configured to perform order based feature engineering matching on the power consumption prediction feature vector and the historical power consumption load data feature vector to obtain a decoding feature vector.
Fig. 5 is a block diagram of a decoding feature generation module in a machine learning based load prediction system according to an embodiment of the present application. As shown in fig. 5, the decoding feature generation module 170 includes: a feature engineering matching factor calculation unit 171 for calculating an order-based feature engineering matching factor between the electricity consumption prediction feature vector and the historical electricity consumption load data feature vector; the weighting unit 172 is configured to weight the power consumption prediction feature vector with the feature engineering matching factor as a weight, so as to obtain a weighted power consumption prediction feature vector; a per-position weighted sum unit 173 for obtaining the decoded feature vector by calculating a per-position weighted sum between the weighted power consumption prediction feature vector and the historical power consumption load data feature vector.
Specifically, in the machine learning-based load prediction system 100, the characteristic engineering matching factor calculation unit 171 is configured to: calculating a characteristic engineering matching factor based on order between the electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector according to the following calculation formula; wherein, the calculation formula is:
Wherein V is 1 Representing the electricity consumption prediction characteristic vector, V 2 Representing the characteristic vector of the historical electricity load data, and T represents the transposition of the vector F Frobenius norms of the matrix are represented, exp (·) is represented by an exponential operation of the matrix, log is represented by a logarithmic function value based on 2, det is represented by a determinant of the matrix, α is represented by a hyper-parameter, w 1 Representing the feature engineering matching factor.
It should be understood that in the technical solution of the present application, the problem of embedding the relative positions between feature vectors of different feature depth dimensions and filter resolutions is converted into an optimization problem by using a feature engineering matching factor based on order, so that an optimization technology is adopted to promote the remote relationship and the probability distribution consistency between the feature vectors. Specifically, firstly, a feature engineering matching factor strategy based on order is designed according to the structure and the attribute of feature vectors with different feature depth dimensions and filter resolutions, and feature values with different categories and dimensions are ordered and grouped according to a certain order rule, so that information redundancy and noise interference in the migration process are reduced. Furthermore, by using an optimization technique, the implicit feature expression of the parameterized model is represented by embedding relative positions between feature vectors with different feature depth dimensions and filter resolutions to achieve a more smooth remote relationship and probability distribution consistency in the high-dimensional feature space. Furthermore, through feature engineering matching analysis based on order, the optimized hidden feature expression is matched with the semantic tag, so that the semantic matching performance of feature vectors with different feature depth dimensions and filter resolutions relative to the semantic tag is realized, the semantic matching capability based on the optimized hidden feature expression is improved, and the semantic matching effect is improved.
In the machine learning based load prediction system 100, the electricity consumption prediction result generation module 180 is configured to pass the decoded feature vector through a decoder to obtain a decoded value, where the decoded value is used to represent a predicted value of the electricity consumption of the current day. Decoding can be regarded as an inverse operation, which restores the decoded feature matrix to the original power consumption prediction value. By inputting the decoded feature vector into the decoder, the learning ability and model parameters of the decoder can be utilized to convert the high-dimensional feature representation into a corresponding power consumption prediction value. The decoder may learn the mapping between the features and the target predicted values to generate the predicted result. Based on the decoded value, a predicted value of the current-day power consumption can be obtained. Therefore, the abstract characteristic information can be converted into a specific prediction result, a practical load prediction value is provided, and a decision maker is helped to make corresponding scheduling and management decisions.
In summary, the machine learning-based load prediction system 100 according to the embodiment of the present application is illustrated, in which the relevant features of the historical power consumption load are obtained by performing feature extraction on the historical weather data, the temperature data, the power consumption data and the date data, the relevant data features affecting the power consumption are obtained by performing feature extraction on the weather data, the temperature data and the date data of the current day, and the extracted features are associated in a high-dimensional space and then passed through a decoder to obtain the decoding value for representing the prediction result of the current day power consumption.
As described above, the machine learning-based load prediction system 100 according to the embodiment of the present application can be implemented in various terminal devices, such as a server or the like for load prediction of machine learning. In one example, the machine learning based load prediction system 100 according to embodiments of the present application may be integrated into a terminal device as one software module and/or hardware module. For example, the machine learning based load prediction system 100 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the machine learning based load prediction system 100 may also be one of a number of hardware modules of the terminal device.
Alternatively, in another example, the machine learning based load prediction system 100 and the terminal device may be separate devices, and the machine learning based load prediction system 100 may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information in a agreed data format.
Exemplary method
Fig. 6 is a flowchart of a machine learning based load prediction method according to an embodiment of the present application. As shown in fig. 6, in the load prediction method based on machine learning, it includes: s110, acquiring historical weather data, temperature data, date data, power consumption data, and weather data, temperature data and date data of the same day; s120, arranging weather data, temperature data and date data of the same day into a power consumption influence parameter matrix, and then obtaining a power consumption prediction feature vector through a first convolution neural network model serving as a feature extractor; s130, passing the historical weather data, temperature data, date data and electricity consumption data through a context encoder comprising an embedded layer to obtain a plurality of historical electricity consumption load related data feature vectors; s140, cascading the historical power consumption load related data feature vectors to obtain a first scale historical power consumption load related data feature vector; s150, two-dimensionally arranging the historical power consumption load related data feature vectors into a historical power consumption load related data semantic feature matrix, and then obtaining a second scale historical power consumption load related data feature vector through a second convolution neural network model comprising a plurality of mixed convolution layers; s160, fusing the first scale historical power consumption load related data feature vector and the second scale historical power consumption load related data feature vector to obtain a historical power consumption load data feature vector; s170, performing order-based feature engineering matching on the electricity consumption prediction feature vector and the historical electricity consumption load data feature vector to obtain a decoding feature vector; and S180, the decoding eigenvector passes through a decoder to obtain a decoding value, wherein the decoding value is used for representing the predicted value of the power consumption of the current day.
Here, it will be understood by those skilled in the art that the specific operations of the respective steps in the above-described machine learning-based load prediction method have been described in detail in the above description of the machine learning-based load prediction system with reference to fig. 1 to 5, and thus, repetitive descriptions thereof will be omitted.
In summary, the machine learning-based load prediction method according to the embodiments of the present application is illustrated, in which the relevant features of the historical power consumption load are obtained by performing feature extraction on the historical weather data, the temperature data, the power consumption data and the date data, then performing feature extraction on the weather data, the temperature data and the date data of the current day to obtain the relevant data features affecting the power consumption, and associating the extracted features in a high-dimensional space and then passing through a decoder to obtain a decoding value for representing the prediction result of the current day power consumption.
Exemplary electronic device
Next, an electronic device according to an embodiment of the present application is described with reference to fig. 7.
Fig. 7 is a block diagram of an electronic device according to an embodiment of the present application.
As shown in fig. 7, the electronic device 10 includes one or more processors 11 and a memory 12.
The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
Memory 12 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like. One or more computer program instructions may be stored on the computer readable storage medium that can be executed by the processor 11 to implement the machine learning based load prediction methods and/or other desired functions of the various embodiments of the present application described above. Various contents such as historical weather data, temperature data, date data, and electricity consumption data, and weather data, temperature data, date data, and the like of the same day may also be stored in the computer-readable storage medium.
In one example, the electronic device 10 may further include: an input device 13 and an output device 14, which are interconnected by a bus system and/or other forms of connection mechanisms (not shown).
The input means 13 may comprise, for example, a keyboard, a mouse, etc.
The output device 14 may output various information to the outside, including the result of the prediction of the current power consumption amount, and the like. The output means 14 may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, etc.
Of course, only some of the components of the electronic device 10 that are relevant to the present application are shown in fig. 7 for simplicity, components such as buses, input/output interfaces, etc. are omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.
Exemplary computer program product and computer readable storage Medium
In addition to the methods and apparatus described above, embodiments of the present application may also be a computer program product comprising computer program instructions which, when executed by a processor, cause the processor to perform the steps in a machine learning based load prediction method according to various embodiments of the present application described in the "exemplary methods" section of this specification.
The computer program product may write program code for performing the operations of embodiments of the present application 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, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present application may also be a computer-readable storage medium, having stored thereon computer program instructions, which when executed by a processor, cause the processor to perform the steps in a machine learning based load prediction method according to various embodiments of the present application described in the above "exemplary method" section of the present specification.
The computer readable storage medium may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
Claims (10)
1. A machine learning based load prediction system, comprising:
The data acquisition module is used for acquiring historical weather data, temperature data, date data, power consumption data, and weather data, temperature data and date data of the same day;
the electricity consumption prediction feature extraction module is used for arranging the weather data, the temperature data and the date data of the current day into an electricity consumption influence parameter matrix and then obtaining electricity consumption prediction feature vectors through a first convolution neural network model serving as a feature extractor;
the historical electricity load data coding module is used for enabling the historical weather data, temperature data, date data and electricity consumption data to pass through a context coder comprising an embedded layer to obtain a plurality of historical electricity load related data feature vectors;
the first scale historical electricity load characteristic generation module is used for cascading the plurality of historical electricity load related data characteristic vectors to obtain a first scale historical electricity load related data characteristic vector;
the second scale historical electricity load characteristic generation module is used for two-dimensionally arranging the plurality of historical electricity load related data characteristic vectors into a historical electricity load related data semantic characteristic matrix and then obtaining a second scale historical electricity load related data characteristic vector through a second convolution neural network model comprising a plurality of mixed convolution layers;
The multi-scale historical power consumption load characteristic fusion module is used for fusing the first-scale historical power consumption load related data characteristic vector and the second-scale historical power consumption load related data characteristic vector to obtain a historical power consumption load data characteristic vector;
the decoding characteristic generation module is used for carrying out order-based characteristic engineering matching on the electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector so as to obtain a decoding characteristic vector;
and the electricity consumption prediction result generation module is used for enabling the decoding feature vector to pass through a decoder to obtain a decoding value, wherein the decoding value is used for representing a predicted value of the electricity consumption of the current day.
2. The machine learning based load prediction system of claim 1, wherein the power consumption prediction feature extraction module is configured to: each layer of the first convolutional neural network model serving as the feature extractor performs input data in the forward transfer process of the layer:
carrying out convolution processing on the input data to obtain a convolution characteristic diagram;
carrying out mean pooling treatment on each feature matrix along the channel dimension on the convolution feature map to obtain a pooled feature map;
Non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map;
the output of the last layer of the first convolutional neural network model is the electricity consumption prediction feature vector, and the input of the first layer of the first convolutional neural network model is the electricity consumption influence parameter matrix.
3. The machine learning based load prediction system of claim 2, wherein the historical electrical load data encoding module comprises:
the word segmentation unit is used for carrying out word segmentation processing on the historical weather data, the historical temperature data, the historical date data and the historical electricity consumption data to obtain a plurality of historical electricity consumption load related words;
the word embedding unit is used for enabling the historical electric load related words to pass through an embedding layer of the context encoder so as to convert each historical electric load related word in the historical electric load related words into a historical electric load related word embedding vector to obtain a sequence of the historical electric load related word embedding vector, wherein the embedding layer uses a learnable embedding matrix to carry out embedding encoding on each historical electric load related word;
and the context semantic coding unit is used for inputting the sequence of the historical electricity load related word embedded vectors into a converter of the context encoder to obtain the plurality of historical electricity load related data feature vectors.
4. A machine learning based load prediction system according to claim 3, characterized in that the context semantic coding unit comprises:
a sequence arrangement subunit, configured to arrange the sequence of the historical electric load related word embedding vector as an input vector;
the vector conversion subunit is used for respectively converting the input vector into a query vector and a key vector through a learning embedding matrix;
a self-attention association matrix generation subunit, configured to calculate a product between the query vector and a transpose vector of the key vector to obtain a self-attention association matrix;
the normalization processing subunit is used for performing normalization processing on the self-attention association matrix to obtain a normalized self-attention association matrix;
an activating subunit, configured to activate the standardized self-attention association matrix input Softmax activating function to obtain a self-attention feature matrix;
and the attention applying subunit is used for multiplying the self-attention characteristic matrix with each historical electric load related word embedded vector in the sequence of the historical electric load related word embedded vectors as a value vector to obtain the plurality of historical electric load related data characteristic vectors.
5. The machine learning based load prediction system of claim 4 wherein the second scale historical electrical load signature generation module is configured to: each layer of the second convolutional neural network model comprising a plurality of mixed convolutional layers respectively carries out input data in the forward transmission process of the layer:
performing multi-scale convolution coding on the input data to obtain a multi-scale feature map;
carrying out mean pooling treatment on each feature matrix along the channel dimension on the multi-scale feature map to obtain a pooled feature map;
non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map;
the input of the first layer of the second convolution neural network model is the historical power consumption load related data semantic feature matrix, and the output of the last layer of the second convolution neural network model is the second scale historical power consumption load related data feature vector.
6. The machine-learning based load prediction system of claim 5, wherein the second scale historical electrical load signature generation module is further configured to:
performing convolution processing on the input data based on a first convolution kernel to obtain a first scale feature map;
Performing convolution processing on the input data based on a second convolution kernel to obtain a second scale feature map, wherein the second convolution kernel is a cavity convolution kernel with first cavity rate;
performing convolution processing on the input data based on a third convolution kernel to obtain a third scale feature map, wherein the third convolution kernel is a cavity convolution kernel with a second cavity rate;
performing convolution processing on the input data based on a fourth convolution kernel to obtain a fourth scale feature map, wherein the fourth convolution kernel is a cavity convolution kernel with a third cavity rate;
and cascading the first scale feature map, the second scale feature map, the third scale feature map and the fourth scale feature map to obtain the multi-scale feature map.
7. The machine-learning-based load prediction system of claim 6, wherein the multi-scale historical electrical load feature fusion module is configured to: fusing the first scale historical electrical load related data feature vector and the second scale historical electrical load related data feature vector using a fusion function to obtain the historical electrical load data feature vector;
Wherein the fusion function is expressed as:
f(X i ,X j )=Relu(W f [θ(X i ),φ(X j )])
wherein W is f ,θ(X i ) And phi (X) j ) All represent the point convolution of the input, relu as the activation function, []Representing the splicing operation, X i Characteristic values, X, representing positions in the first scale historical electrical load related data characteristic vector j Characteristic values representing respective positions in the second-scale historical electric load related data characteristic vector, f (X) i ,X j ) And representing the historical electricity load data characteristic vector.
8. The machine learning based load prediction system of claim 7, wherein the decode feature generation module comprises:
the characteristic engineering matching factor calculation unit is used for calculating characteristic engineering matching factors based on order between the electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector;
the weighting unit is used for weighting the electricity prediction feature vector by taking the feature engineering matching factor as a weight so as to obtain a weighted electricity prediction feature vector;
and the per-position weighted sum unit is used for calculating a per-position weighted sum between the weighted electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector to obtain the decoding characteristic vector.
9. The machine learning based load prediction system of claim 8, wherein the feature engineering matching factor calculation unit is configured to: calculating a characteristic engineering matching factor based on order between the electricity consumption prediction characteristic vector and the historical electricity consumption load data characteristic vector according to the following calculation formula;
wherein, the calculation formula is:
wherein V is 1 Representing the electricity consumption prediction characteristic vector, V 2 Representing the characteristic vector of the historical electricity load data, and T represents the transposition of the vector F Frobenius norms of the matrix are represented, exp (·) is represented by an exponential operation of the matrix, log is represented by a logarithmic function value based on 2, det is represented by a determinant of the matrix, α is represented by a hyper-parameter, w 1 Representing the feature engineering matching factor.
10. A machine learning based load prediction method, comprising:
acquiring historical weather data, temperature data, date data, electricity consumption data and weather data, temperature data and date data of the same day;
the weather data, the temperature data and the date data of the current day are arranged into an electricity consumption influence parameter matrix, and then an electricity consumption prediction feature vector is obtained through a first convolution neural network model serving as a feature extractor;
Passing the historical weather data, temperature data, date data and electricity consumption data through a context encoder comprising an embedded layer to obtain a plurality of historical electricity consumption load related data feature vectors;
cascading the plurality of historical electricity load related data feature vectors to obtain a first-scale historical electricity load related data feature vector;
the historical power consumption load related data feature vectors are two-dimensionally arranged into a historical power consumption load related data semantic feature matrix, and then a second scale historical power consumption load related data feature vector is obtained through a second convolution neural network model comprising a plurality of mixed convolution layers;
fusing the first-scale historical power consumption load related data feature vector and the second-scale historical power consumption load related data feature vector to obtain a historical power consumption load data feature vector;
performing order-based feature engineering matching on the electricity consumption prediction feature vector and the historical electricity consumption load data feature vector to obtain a decoding feature vector;
and the decoding eigenvector is passed through a decoder to obtain a decoding value, wherein the decoding value is used for representing the predicted value of the current power consumption.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311567701.9A CN117744855A (en) | 2023-11-22 | 2023-11-22 | Load prediction system and method based on machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311567701.9A CN117744855A (en) | 2023-11-22 | 2023-11-22 | Load prediction system and method based on machine learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117744855A true CN117744855A (en) | 2024-03-22 |
Family
ID=90251673
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311567701.9A Pending CN117744855A (en) | 2023-11-22 | 2023-11-22 | Load prediction system and method based on machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117744855A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117977587A (en) * | 2024-04-02 | 2024-05-03 | 南京鼎研电力科技有限公司 | Power load prediction system and method based on deep neural network |
-
2023
- 2023-11-22 CN CN202311567701.9A patent/CN117744855A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117977587A (en) * | 2024-04-02 | 2024-05-03 | 南京鼎研电力科技有限公司 | Power load prediction system and method based on deep neural network |
CN117977587B (en) * | 2024-04-02 | 2024-06-07 | 南京鼎研电力科技有限公司 | Power load prediction system and method based on deep neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Lindemann et al. | A survey on long short-term memory networks for time series prediction | |
CN115796173B (en) | Data processing method and system for supervising reporting requirements | |
Fan et al. | A novel deep generative modeling-based data augmentation strategy for improving short-term building energy predictions | |
CN117744855A (en) | Load prediction system and method based on machine learning | |
Sun et al. | Two-channel attention mechanism fusion model of stock price prediction based on CNN-LSTM | |
Kim et al. | Short-term electrical load forecasting with multidimensional feature extraction | |
CN115878330A (en) | Thread operation control method and system | |
CN116151604A (en) | Office system flow analysis system and method under web environment | |
CN115759658A (en) | Enterprise energy consumption data management system suitable for smart city | |
CN115373813A (en) | Scheduling method and system based on GPU virtualization in cloud computing environment and electronic equipment | |
CN116307624A (en) | Resource scheduling method and system of ERP system | |
CN114118570A (en) | Service data prediction method and device, electronic equipment and storage medium | |
CN117875638A (en) | Big data-based power supply demand distribution method and system | |
Luo et al. | Multi-task prediction model based on ConvLSTM and encoder-decoder | |
Jiang et al. | Mobile traffic prediction in consumer applications: a multimodal deep learning approach | |
CN117077955A (en) | Energy management system and method for optical storage grid-connected power generation system | |
CN116186295B (en) | Attention-based knowledge graph link prediction method, attention-based knowledge graph link prediction device, attention-based knowledge graph link prediction equipment and attention-based knowledge graph link prediction medium | |
CN117972355A (en) | Time sequence data prediction method, system, equipment and medium | |
CN113554149B (en) | Neural network processing unit NPU, neural network processing method and device | |
CN115238009A (en) | Metadata management method, device and equipment based on blood vessel margin analysis and storage medium | |
CN114726870A (en) | Hybrid cloud resource arrangement method and system based on visual dragging and electronic equipment | |
Nutakki et al. | Optimizing home energy management: Robust and efficient solutions powered by attention networks | |
CN113420869A (en) | Translation method based on omnidirectional attention and related equipment thereof | |
CN113760497A (en) | Scheduling task configuration method and device | |
CN116561696B (en) | Multi-dimensional user adjustable load rapid aggregation method and system thereof |
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 |