Energy demand prediction system and method based on data fusion and deep learning

Abstract

The invention provides an energy demand prediction system and method based on data fusion and deep learning, and the system comprises a data module which is used for collecting and preprocessing multi-source energy data; the fusion module is used for carrying out data fusion on the processed current multi-source energy data; the initial prediction model construction module is used for inputting the processed historical multi-source energy data and the fused multi-source energy data into a deep learning model for training to obtain a trained initial prediction model; the emergency quantification module is used for constructing an emergency influence quantification model based on the historical emergency data; the embedding module is used for embedding the emergency influence quantitative model into the initial prediction model to obtain a final energy demand prediction model; and the prediction result output and explanation module is used for inputting the new energy data into the energy demand prediction model, generating a corresponding prediction result, and explaining and visually displaying the prediction result.

Description

Technical Field

[0001] The present invention relates to the technical field of energy resources, and particularly to an energy demand prediction system and method based on data fusion and deep learning. Background Art

[0002] Energy resources are an important material guarantee for the sustainable development of the economy and society. Energy demand prediction directly affects important decisions on future social and economic development and energy supply, and these decisions will in turn have a significant impact on future energy development. Therefore, it is particularly important to scientifically and reasonably predict energy demand.

[0003] Currently, energy demand prediction is usually based on traditional statistical models and physical models. These methods consider various influencing factors such as economic growth, population development, industrial structure, and energy consumption structure during the prediction process, and to a certain extent improve the prediction accuracy. However, these methods still have obvious limitations:

[0004] 1. Limited information sources: The information reflected by traditional prediction models is limited, lacking the cognitive ability to integrate various types of prediction information, increasing the uncertainty of the impact of a single information source on energy demand prediction during the prediction process;

[0005] 2. Insufficient processing of non-linear relationships: The energy system is a complex non-linear system, and its demand is affected by the interweaving of multiple factors. Traditional models are unable to cope when dealing with these complex non-linear relationships;

[0006] 3. Difficulty in quantifying the impact of emergencies: The short-term volatility of energy demand is often caused by sudden irregular events such as financial crises, wars, natural disasters, and coups. These irregular events are characterized by non-quantifiable and uncertain characteristics and often cannot be incorporated into traditional mathematical models, resulting in difficulty in fundamentally improving the prediction effect.

[0007] In order to overcome the limitations of traditional prediction methods, multi-source data fusion technology has gradually been introduced into the field of energy demand prediction. The basic principle of multi-source data fusion technology is to make full use of the information provided by multiple information sources. By reasonably controlling and using these information sources and the information they provide, redundant or complementary information from multiple information sources in space or time is combined according to a certain criterion to obtain a consistent interpretation or description of the object to be measured. This method can significantly improve the performance of information systems and reduce the impact of a single information source on the prediction result. However, although multi-source data fusion technology shows great potential in improving prediction accuracy, its application in the field of energy demand prediction is still in its infancy and no mature and systematic solution has been formed.

[0008] In recent years, deep learning algorithms have achieved remarkable results in various fields by virtue of their powerful data processing and pattern recognition capabilities. In the aspect of energy consumption prediction, deep learning models can automatically extract key features from a large amount of historical energy data, capture complex non-linear relationships, and effectively process high-dimensional data and large-scale data sets. In addition, time series models such as recurrent neural networks (RNNs) and long short-term memory networks (LSTMs) in deep learning can effectively capture long-term dependencies in time series, further improving the prediction accuracy. Although deep learning shows great potential in energy consumption prediction, systems and methods that combine it with multi-source data fusion technology for energy demand prediction are yet to be developed.

[0009] In summary, existing energy demand prediction methods have obvious limitations in aspects such as information source integration, non-linear relationship processing, and quantification of the impact of emergencies. Multi-source data fusion technology and deep learning algorithms provide new ideas and methods to overcome these limitations. Therefore, it is of great theoretical and practical significance to develop an energy demand prediction system and method based on multi-source data fusion and deep learning. Summary of the Invention

[0010] The purpose of the present invention is to provide an energy demand prediction system and method based on data fusion and deep learning, aiming to solve the above problems in the prior art.

[0011] An embodiment of the present invention provides an energy demand prediction system based on data fusion and deep learning, including:

[0012] A data module, connected to a fusion module, an initial prediction model construction module, and an emergency quantification module, for collecting multi-source energy data, preprocessing the multi-source energy data, and transmitting the processed energy data to the fusion module, the initial prediction model construction module, and the emergency quantification module; wherein, the multi-source energy data includes historical multi-source energy data and current multi-source energy data;

[0013] A fusion module, connected to the data module and the initial prediction model construction module, for performing data fusion on the processed current multi-source energy data, and transmitting the fused multi-source energy data to the initial prediction model construction module;

[0014] An initial prediction model construction module, connected to the data module, the fusion module, and an embedding module, for receiving the processed historical multi-source energy data and the fused multi-source energy data, inputting the processed historical multi-source energy data and the fused multi-source energy data into a deep learning model for training to obtain a trained initial prediction model, and transmitting the initial prediction model to the embedding module;

[0015] An emergency quantification module, connected to the data module and the embedding module, is used to construct an emergency impact quantification model based on historical emergency data in the processed historical multi-source energy data, and transmit the emergency impact quantification model to the embedding module;

[0016] An embedding module, connected to the initial prediction model construction module, the emergency quantification module, and the prediction result output and interpretation module, is used to embed the emergency impact quantification model into the initial prediction model to obtain a final energy demand prediction model, and send the energy demand prediction model to the prediction result output and interpretation module;

[0017] A prediction result output and interpretation module, connected to the embedding module, is used to input new energy data into the energy demand prediction model, generate corresponding prediction results, and interpret and visually display the prediction results.

[0018] An embodiment of the present invention provides an energy demand prediction method based on data fusion and deep learning, including:

[0019] Collect multi-source energy data through a data module, preprocess the multi-source energy data, and transmit the processed energy data to a fusion module, an initial prediction model construction module, and an emergency quantification module; wherein, the multi-source energy data includes historical multi-source energy data and current multi-source energy data;

[0020] The fusion module performs data fusion on the processed current multi-source energy data and transmits the fused multi-source energy data to the initial prediction model construction module;

[0021] The initial prediction model construction module receives the processed historical multi-source energy data and the fused multi-source energy data, inputs the processed historical multi-source energy data and the fused multi-source energy data into a deep learning model for training to obtain a trained initial prediction model, and transmits the initial prediction model to the embedding module;

[0022] The emergency quantification module constructs an emergency impact quantification model based on historical emergency data in the processed historical multi-source energy data and transmits the emergency impact quantification model to the embedding module;

[0023] The embedding module embeds the emergency impact quantification model into the initial prediction model to obtain a final energy demand prediction model, and sends the energy demand prediction model to the prediction result output and interpretation module;

[0024] The new energy data is input into the energy demand prediction model through the prediction result output and explanation module, and the corresponding prediction result is generated, and the prediction result is explained and visually displayed.

[0025] An embodiment of the present invention further provides an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the computer program is executed by the processor, the steps of the above-mentioned energy demand prediction method based on data fusion and deep learning are implemented.

[0026] An embodiment of the present invention further provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by a processor, the steps of the above-mentioned energy demand prediction method based on data fusion and deep learning are implemented.

[0027] The adoption of the embodiment of the present invention may include the following beneficial effects: The embodiment of the present invention aims to provide an energy demand prediction system and method based on multi-source data fusion and deep learning to overcome the limitations of existing energy demand prediction methods in aspects such as information source integration, non-linear relationship processing, and quantification of the impact of emergencies. The system and method can make full use of the information provided by multiple information sources, automatically extract key features through deep learning algorithms, capture complex non-linear relationships, and effectively process high-dimensional data and large-scale data sets, thereby improving the accuracy and robustness of energy demand prediction. Description of the Drawings

[0028] In order to more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in this specification. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0029] Figure 1 It is a schematic diagram of the energy demand prediction system based on data fusion and deep learning according to an embodiment of the present invention;

[0030] Figure 2 It is a flowchart of the energy demand prediction method based on data fusion and deep learning according to an embodiment of the present invention. Detailed Embodiments

[0031] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will clearly and completely describe the technical solutions in one or more embodiments of this specification in conjunction with the accompanying drawings in one or more embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this document.

[0032] System Embodiment

[0033] According to an embodiment of the present invention, there is provided an energy demand prediction system based on data fusion and deep learning. Figure 1 It is a schematic diagram of the energy demand prediction system based on data fusion and deep learning in an embodiment of the present invention. As Figure 1 shown, the energy demand prediction system based on data fusion and deep learning according to an embodiment of the present invention specifically includes:

[0034] A data module 10, connected to a fusion module, an initial prediction model construction module, and an emergency quantification module, is used to collect multi-source energy data, preprocess the multi-source energy data, and transmit the processed energy data to the fusion module, the initial prediction model construction module, and the emergency quantification module; wherein, the multi-source energy data includes historical multi-source energy data and current multi-source energy data.

[0035] A fusion module 11, connected to the data module and the initial prediction model construction module, is used to perform data fusion on the processed current multi-source energy data and transmit the fused multi-source energy data to the initial prediction model construction module, and specifically includes:

[0036] An association and matching unit, connected to a fusion strategy unit, is used to associate and match the processed current multi-source energy data according to the time stamp and geographical location of the data, and construct a corresponding data relationship model.

[0037] A fusion strategy unit, connected to the association and matching unit, is used to assign corresponding weights to different data sources in the processed current multi-source energy data according to the data nature, and perform data fusion on the multi-source energy data with assigned weights by using a multi-type data strategy.

[0038] Among them, the multi-type data strategy includes a weighted average algorithm, a Kalman filter algorithm, and a Bayesian network model.

[0039] The initial prediction model construction module 12, connected to the data module, the fusion module, and the embedding module, is used to receive the processed historical multi-source energy data and the fused multi-source energy data, input the processed historical multi-source energy data and the fused multi-source energy data into a deep learning model for training to obtain a trained initial prediction model, and transmit the initial prediction model to the embedding module;

[0040] The emergency quantification module 13, connected to the data module and the embedding module, is used to construct an emergency impact quantification model based on the historical emergency data in the processed historical multi-source energy data and transmit the emergency impact quantification model to the embedding module. Specifically, it is used for:

[0041] Analyze the historical emergency data in the processed historical multi-source energy data and its impact on energy demand, and construct an emergency impact quantification model using an expert prediction system or a machine learning algorithm according to the analysis results;

[0042] The embedding module 14, connected to the initial prediction model construction module, the emergency quantification module, and the prediction result output and interpretation module, is used to embed the emergency impact quantification model into the initial prediction model to obtain a final energy demand prediction model, and send the energy demand prediction model to the prediction result output and interpretation module;

[0043] The prediction result output and interpretation module 15, connected to the embedding module, is used to input new energy data into the energy demand prediction model, generate corresponding prediction results, and interpret and visually display the prediction results;

[0044] The system further includes:

[0045] The verification module, connected to the prediction result output and interpretation module and the continuous optimization module, is used to compare the prediction results with the actual energy demand data, verify the prediction performance of the system according to the comparison results, generate verification results, and transmit the verification results to the continuous optimization module;

[0046] The continuous optimization module, connected to the verification module, is used to collect user feedback and receive the verification results, and continuously optimize the energy demand prediction model in the system based on the user feedback and the verification results.

[0047] The above technical solutions of the embodiments of the present invention will be described in detail below in combination with the specific situation of the energy demand prediction system based on data fusion and deep learning of the embodiments of the present invention.

[0048] The system proposed in the embodiments of the present invention includes the following modules:

[0049] 1. Data Acquisition and Preprocessing Module

[0050] This module is responsible for collecting data from different information sources, including but not limited to economic growth data, population development data, industrial structure data, energy consumption structure data, meteorological data, policy data, etc. At the same time, preprocess the collected data, including data cleaning, data transformation, data normalization and other steps to ensure the quality and consistency of the data.

[0051] Among them, the data acquisition and preprocessing module specifically includes:

[0052] Data acquisition unit, which is used to obtain macroeconomic indicators such as economic growth data, population development data, and industrial structure data from official channels, obtain energy consumption structure data, energy price data, etc. from energy enterprises, industry associations and other channels, obtain meteorological data from meteorological departments, and obtain relevant policy data from policy-making institutions;

[0053] Data preprocessing unit, which is used to clean, transform and normalize the collected data.

[0054] 2. Multi-source Data Fusion Module

[0055] This module uses advanced multi-source data fusion technology to fuse data from different information sources. By considering the correlation, complementarity and redundancy between data, select appropriate fusion strategies and methods, such as weighted average, Kalman filter, Bayesian network, etc., to obtain a consistent interpretation or description of the measured object.

[0056] Among them, the multi-source data fusion module specifically includes:

[0057] Data association and matching unit, which is used to associate and match data from different sources according to information such as the timestamp and geographical location of the data;

[0058] Data fusion strategy unit, which is used to assign different weights to different data sources according to the reliability and importance of the data, and use methods such as weighted average, Kalman filter, Bayesian network, etc. for data fusion.

[0059] 3. Deep Learning Prediction Model Construction Module

[0060] This module constructs a prediction model based on deep learning algorithms. First, select a suitable deep learning architecture according to the characteristics and requirements of energy demand prediction, such as recurrent neural network (RNN), long short-term memory network (LSTM), convolutional neural network (CNN), etc.; then, use historical energy data and fused multi-source data to train and optimize the model to improve the prediction accuracy and generalization ability of the model.

[0061] Among them, the deep learning prediction model construction module specifically includes:

[0062] The model selection unit is used to select a suitable deep learning architecture according to the characteristics and requirements of energy demand prediction, such as recurrent neural network (RNN), long short-term memory network (LSTM), convolutional neural network (CNN), etc.

[0063] The model training and optimization unit is used to train the model with the fused data, and adopts techniques such as cross-validation, regularization, and early stopping to prevent the model from overfitting, and optimizes the training effect of the model by adjusting hyperparameters such as learning rate and batch size.

[0064] 4. Prediction result output and interpretation module

[0065] This module is responsible for interpreting and visualizing the output results of the deep learning prediction model. By generating prediction reports, charts, curves, etc., it intuitively displays the prediction results and trends of energy demand. At the same time, it analyzes and evaluates the uncertainty and error of the prediction results to provide a reliable prediction basis and decision support.

[0066] Among them, the prediction result output and interpretation module specifically includes:

[0067] The prediction result generation unit is used to apply the trained deep learning model to new data input, generate the prediction results of energy demand, and post-process the prediction results.

[0068] The result interpretation and visualization unit is used to generate forms such as prediction reports, charts, and curves to intuitively display the prediction results and trends of energy demand, and analyze and evaluate the uncertainty and error of the prediction results.

[0069] 5. Emergency impact quantification module

[0070] Aiming at the problem that it is difficult to quantify the impact of emergencies on energy demand, this module uses an expert prediction system or machine learning algorithm to predict and quantify emergencies. By collecting and analyzing historical emergency data and their impact on energy demand, an emergency impact quantification model is established. During the prediction process, the impact of emergencies is incorporated into the prediction model to improve the accuracy and reliability of the prediction results.

[0071] Among them, the emergency impact quantification module specifically includes:

[0072] The emergency prediction unit is used to collect and analyze historical emergency data and their impact on energy demand, and uses an expert prediction system or machine learning algorithm to predict and quantify emergencies.

[0073] An impact incorporation prediction model unit is used to incorporate the impact of emergencies into a deep learning prediction model, adjust the parameters and structure of the model to adapt to the impact of emergencies, update emergency data in real time, and dynamically adjust the prediction results.

[0074] Specifically, the embodiments of the present invention include the following:

[0075] I. System architecture

[0076] The system proposed in the embodiments of the present invention mainly includes a data collection and preprocessing module, a multi-source data fusion module, a deep learning prediction model construction module, a prediction result output and interpretation module, and an emergency impact quantification module. Each module is connected and interacted through data interfaces and communication protocols to form a complete energy demand prediction system.

[0077] II. Data collection and preprocessing

[0078] ① Data collection:

[0079] Obtain macroeconomic indicators such as economic growth data, population development data, and industrial structure data from official channels.

[0080] Obtain energy consumption structure data, energy price data, etc. from energy enterprises, industry associations and other channels.

[0081] Obtain meteorological data from the meteorological department, including temperature, humidity, wind speed, precipitation, etc.

[0082] Obtain relevant policy data from policy-making institutions, such as energy policies, environmental protection policies, etc.

[0083] ② Data preprocessing:

[0084] Clean the collected data to remove duplicate, incorrect, and missing data.

[0085] Convert the data, such as converting non-numerical data to numerical data, and perform normalization processing on the data.

[0086] Store the processed data in the database for subsequent use.

[0087] III. Multi-source data fusion

[0088] ① Data association and matching:

[0089] According to information such as the timestamp and geographical location of the data, associate and match data from different sources.

[0090] Construct a data model to describe the relationships and dependencies between different data sources.

[0091] ② Data fusion strategy:

[0092] Assign different weights to different data sources according to the reliability and importance of the data.

[0093] Adopt methods such as weighted average, Kalman filtering, and Bayesian network for data fusion to obtain a consistent interpretation or description of the object under test.

[0094] IV. Construction of Deep Learning Prediction Model

[0095] ① Model Selection:

[0096] Select a suitable deep learning architecture according to the characteristics and requirements of energy demand prediction, such as recurrent neural network (RNN), long short-term memory network (LSTM), convolutional neural network (CNN), etc.

[0097] Consider factors such as model complexity, training time, and prediction accuracy for model optimization.

[0098] ② Model Training and Optimization:

[0099] Train the model using historical energy data and the fused multi-source data.

[0100] Adopt techniques such as cross-validation, regularization, and early stopping to prevent model overfitting.

[0101] Optimize the training effect of the model by adjusting hyperparameters such as learning rate and batch size.

[0102] V. Prediction Result Output and Interpretation

[0103] ① Prediction Result Generation:

[0104] Apply the trained deep learning model to new data inputs to generate prediction results for energy demand.

[0105] Post-process the prediction results, such as smoothing and outlier detection.

[0106] ② Result Interpretation and Visualization:

[0107] Generate prediction reports, charts, curves, etc. to visually display the prediction results and trends of energy demand.

[0108] Analyze and evaluate the uncertainty and error of the prediction results to provide a reliable prediction basis and decision support.

[0109] VI. Quantification of the Impact of Emergencies

[0110] ① Emergency Prediction:

[0111] Collect and analyze historical emergency data and their impact on energy demand.

[0112] Use an expert prediction system or machine learning algorithm to predict and quantify emergencies.

[0113] ② Influence incorporated into the prediction model:

[0114] Incorporate the impact of emergencies into the deep learning prediction model, and adjust the parameters and structure of the model to adapt to the impact of emergencies.

[0115] Update emergency data in real time and dynamically adjust the prediction results.

[0116] VII. System Testing and Verification

[0117] ① Data testing:

[0118] Use historical data to test the system and verify the prediction accuracy and robustness of the system.

[0119] Compare the system prediction results with the actual energy demand data to evaluate the prediction performance of the system.

[0120] ② User feedback:

[0121] Collect user feedback and continuously optimize and improve the system.

[0122] Adjust the parameters and structure of the prediction model according to user needs to improve the practicality of the system and the user experience.

[0123] Through the above specific implementation manners, the embodiments of the present invention provide an energy demand prediction system based on multi-source data fusion and deep learning. The system can make full use of the information provided by multiple information sources, capture complex non-linear relationships, improve the prediction accuracy and robustness, and provide reliable decision-making basis and support for governments, enterprises and research institutions.

[0124] Method Embodiment

[0125] According to an embodiment of the present invention, there is provided an energy demand prediction method based on data fusion and deep learning. Figure 2 It is a flowchart of the energy demand prediction method based on data fusion and deep learning according to an embodiment of the present invention. As Figure 2 shown, the energy demand prediction method based on data fusion and deep learning according to an embodiment of the present invention specifically includes:

[0126] Step S201, collect multi-source energy data through a data module, preprocess the multi-source energy data, and transmit the processed energy data to a fusion module, an initial prediction model construction module, and an emergency quantification module; wherein, the multi-source energy data includes historical multi-source energy data and current multi-source energy data.

[0127] Step S202: The processed current multi-source energy data is fused by the fusion module, and the fused multi-source energy data is transmitted to the initial prediction model construction module, which specifically includes:

[0128] The associated and matched unit in the fusion module associates and matches the processed current multi-source energy data according to the timestamp and geographical location of the data, and constructs a corresponding data relationship model;

[0129] The fusion strategy unit in the fusion module assigns corresponding weights to different data sources in the processed current multi-source energy data according to the data nature, and performs data fusion on the multi-source energy data with assigned weights by using multi-type data strategies;

[0130] Among them, the multi-type data strategies include weighted average algorithm, Kalman filtering algorithm, and Bayesian network model;

[0131] Step S203: The initial prediction model construction module receives the processed historical multi-source energy data and the fused multi-source energy data, inputs the processed historical multi-source energy data and the fused multi-source energy data into the deep learning model for training to obtain a trained initial prediction model, and transmits the initial prediction model to the embedding module;

[0132] Step S204: The emergency quantification module constructs an emergency impact quantification model based on the historical emergency data in the processed historical multi-source energy data, and transmits the emergency impact quantification model to the embedding module, which specifically includes:

[0133] The emergency quantification module analyzes the historical emergency data in the processed historical multi-source energy data and its impact on energy demand, and constructs an emergency impact quantification model by using an expert prediction system or a machine learning algorithm according to the analysis results;

[0134] Step S205: The embedding module embeds the emergency impact quantification model into the initial prediction model to obtain a final energy demand prediction model, and sends the energy demand prediction model to the prediction result output and interpretation module;

[0135] Step S206: The prediction result output and interpretation module inputs new energy data into the energy demand prediction model, generates corresponding prediction results, and interprets and visually displays the prediction results;

[0136] The method further includes:

[0137] The prediction result is compared with the actual energy demand data through a verification module, and the prediction performance of the system is verified according to the comparison result to generate a verification result, and the verification result is transmitted to a continuous optimization module;

[0138] The continuous optimization module collects user feedback and receives the verification result, and continuously optimizes the energy demand prediction model in the system based on the user feedback and the verification result.

[0139] The embodiment of the present invention is a method embodiment corresponding to the above system embodiment. The specific operations of each step can be understood with reference to the description of the system embodiment, and will not be elaborated here.

[0140] In summary, the embodiment of the present invention specifically includes the following beneficial effects:

[0141] 1. Improve prediction accuracy: By applying multi-source data fusion and deep learning algorithms, it is possible to fully utilize the information provided by multiple information sources, capture complex non-linear relationships, and improve the accuracy and robustness of prediction.

[0142] 2. Enhance generalization ability: The deep learning model can automatically extract key features from a large amount of historical data and adapt to energy demand prediction under different scenarios and conditions.

[0143] 3. Achieve real-time prediction: By collecting and processing data in real time, and using efficient deep learning algorithms, it is possible to achieve real-time prediction and dynamic adjustment of energy demand.

[0144] 4. Provide decision-making support: The prediction result can provide a reliable decision-making basis for the government, enterprises and research institutions, and contribute to scientific production capacity planning, formulation of energy development plans and response to emergencies, etc.

[0145] Device Embodiment 1

[0146] The embodiment of the present invention provides an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps as described in the method embodiment.

[0147] Device Embodiment 2

[0148] The embodiment of the present invention provides a computer-readable storage medium, on which an implementation program for information transmission is stored. When the program is executed by a processor, it implements the steps as described in the method embodiment.

[0149] The computer-readable storage medium described in this embodiment includes, but is not limited to: ROM, RAM, magnetic disk or optical disk, etc.

[0150] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An energy demand prediction system based on data fusion and deep learning, characterized in that Including: A data module, connected to the fusion module, the initial prediction model construction module, and the emergency event quantification module, for collecting multi-source energy data, preprocessing the multi-source energy data, and transmitting the processed energy data to the fusion module, the initial prediction model construction module, and the emergency event quantification module; wherein, the multi-source energy data includes historical multi-source energy data and current multi-source energy data; A fusion module, connected to the data module and the initial prediction model construction module, for performing data fusion on the processed current multi-source energy data, and transmitting the fused multi-source energy data to the initial prediction model construction module; An initial prediction model construction module, connected to the data module, the fusion module, and the embedding module, for receiving the processed historical multi-source energy data and the fused multi-source energy data, inputting the processed historical multi-source energy data and the fused multi-source energy data into a deep learning model for training to obtain a trained initial prediction model, and transmitting the initial prediction model to the embedding module; An emergency event quantification module, connected to the data module and the embedding module, for constructing an emergency event impact quantification model based on the historical emergency event data in the processed historical multi-source energy data, and transmitting the emergency event impact quantification model to the embedding module; An embedding module, connected to the initial prediction model construction module, the emergency event quantification module, and the prediction result output and interpretation module, for embedding the emergency event impact quantification model into the initial prediction model to obtain a final energy demand prediction model, and sending the energy demand prediction model to the prediction result output and interpretation module; A prediction result output and interpretation module, connected to the embedding module, for inputting new energy data into the energy demand prediction model, generating a corresponding prediction result, and interpreting and visually displaying the prediction result.

2. The system according to claim 1, wherein The system further includes: A verification module, connected to the prediction result output and interpretation module and the continuous optimization module, for comparing the prediction result with the actual energy demand data, verifying the prediction performance of the system according to the comparison result, generating a verification result, and transmitting the verification result to the continuous optimization module; A continuous optimization module, connected to the verification module, for collecting user feedback and receiving the verification result, and continuously optimizing the energy demand prediction model in the system based on the user feedback and the verification result.

3. The system according to claim 1, characterized in that The fusion module specifically includes: An association and matching unit, connected to the fusion strategy unit, for associating and matching the processed current multi-source energy data according to the time stamp and geographical location of the data, and constructing a corresponding data relationship model; A fusion strategy unit, connected to the association and matching unit, for assigning corresponding weights to different data sources in the processed current multi-source energy data according to the data nature, and performing data fusion on the multi-source energy data with assigned weights by using a multi-type data strategy; Wherein, the multi-type data strategy includes a weighted average algorithm, a Kalman filter algorithm, and a Bayesian network model.

4. The system according to claim 1, wherein The emergency quantification module is specifically used for: Analyze the historical emergency data and its impact on energy demand in the processed historical multi-source energy data, and construct an emergency impact quantification model using an expert prediction system or a machine learning algorithm according to the analysis results.

5. A method for predicting energy demand based on data fusion and deep learning, characterized in that Including: Collect multi-source energy data through the data module, preprocess the multi-source energy data, and transmit the processed energy data to the fusion module, the initial prediction model construction module, and the emergency quantification module; wherein, the multi-source energy data includes historical multi-source energy data and current multi-source energy data; Fuse the processed current multi-source energy data through the fusion module, and transmit the fused multi-source energy data to the initial prediction model construction module; Receive the processed historical multi-source energy data and the fused multi-source energy data through the initial prediction model construction module, input the processed historical multi-source energy data and the fused multi-source energy data into a deep learning model for training to obtain a trained initial prediction model, and transmit the initial prediction model to the embedding module; Construct an emergency impact quantification model based on the historical emergency data in the processed historical multi-source energy data through the emergency quantification module, and transmit the emergency impact quantification model to the embedding module; Embed the emergency impact quantification model into the initial prediction model through the embedding module to obtain a final energy demand prediction model, and send the energy demand prediction model to the prediction result output and explanation module; Input new energy data into the energy demand prediction model through the prediction result output and explanation module to generate corresponding prediction results, and explain and visually display the prediction results.

6. The method according to claim 5, wherein The method further includes: Compare the prediction results with the actual energy demand data through the verification module, verify the prediction performance of the system according to the comparison results, generate verification results, and transmit the verification results to the continuous optimization module; Collect user feedback and receive the verification results through the continuous optimization module, and continuously optimize the energy demand prediction model in the system based on the user feedback and the verification results.

7. The method according to claim 5, wherein Specifically, the fusion of the processed current multi-source energy data by the fusion module includes: Associate and match the processed current multi-source energy data according to the time stamp and geographical location of the data through the association and matching unit in the fusion module to construct a corresponding data relationship model; Allocate corresponding weights to different data sources in the processed current multi-source energy data according to the data nature through the fusion strategy unit in the fusion module, and perform data fusion on the weighted multi-source energy data using a multi-type data strategy; Wherein, the multi-type data strategy includes a weighted average algorithm, a Kalman filter algorithm, and a Bayesian network model.

8. The method according to claim 5, characterized in that, Specifically, constructing an emergency impact quantification model based on the historical emergency data in the processed historical multi-source energy data through the emergency quantification module includes: The historical emergency data in the processed historical multi-source energy data and its impact degree on energy demand are analyzed by an emergency quantification module, and an emergency impact quantification model is constructed by using an expert prediction system or a machine learning algorithm according to the analysis results.

9. An electronic device, characterized in that, Including: A memory, a processor, and a computer program stored on the memory and executable on the processor, and when the computer program is executed by the processor, the steps of the energy demand prediction method based on data fusion and deep learning according to any one of claims 5-8 are implemented.

10. A computer-readable storage medium, characterized in that, An implementation program for information transmission is stored on the computer-readable storage medium, and when the program is executed by the processor, the steps of the energy demand prediction method based on data fusion and deep learning according to any one of claims 5-8 are implemented.

Images