CN118114816A - Rapid prediction method for building energy consumption - Google Patents

Rapid prediction method for building energy consumption Download PDF

Info

Publication number
CN118114816A
CN118114816A CN202410181691.3A CN202410181691A CN118114816A CN 118114816 A CN118114816 A CN 118114816A CN 202410181691 A CN202410181691 A CN 202410181691A CN 118114816 A CN118114816 A CN 118114816A
Authority
CN
China
Prior art keywords
energy consumption
data
building
prediction
energy
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
Application number
CN202410181691.3A
Other languages
Chinese (zh)
Inventor
徐燊
崔国游
吴玉杰
李辉
沈念俊
韩昀松
李高梅
何秋国
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN202410181691.3A priority Critical patent/CN118114816A/en
Publication of CN118114816A publication Critical patent/CN118114816A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/10—Pre-processing; Data cleansing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G06N3/0455—Auto-encoder networks; Encoder-decoder networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0475—Generative networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/094—Adversarial learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/096—Transfer learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/08—Construction

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Business, Economics & Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Economics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Primary Health Care (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention relates to the technical field of building energy consumption prediction, and discloses a rapid prediction method of building energy consumption, which comprises the steps of acquiring environmental parameter data in a building through deployment of a sensor network, wherein the acquired data comprise real-time data or historical data; the method comprises the steps of carrying out data preprocessing on collected environment parameter data, training a building energy consumption prediction model by using the preprocessed environment parameter data and corresponding energy consumption data through a deep learning algorithm, building a complex relation between environment parameters and energy consumption through model training, carrying out rapid prediction on building energy consumption by combining the real-time collected environment parameter data through the trained deep learning model, and generating an effective energy management strategy according to an energy consumption prediction result and user requirements. According to the building energy consumption rapid prediction method, the complex nonlinear relation can be processed by collecting the environmental parameter data of the building and utilizing the deep learning algorithm to perform model training and prediction, so that the rapid and accurate prediction of the building energy consumption is realized.

Description

Rapid prediction method for building energy consumption
Technical Field
The invention relates to the technical field of building energy consumption prediction, in particular to a rapid building energy consumption prediction method.
Background
During the use period of the building, energy consumption can be generated due to the requirements of heating, cooling, illumination and the like for using the electric appliances. The task of building energy consumption prediction is researched, the more accurate prediction of the energy consumption condition of a specific building can be made, and the method is an important precondition for developing a green building; the energy can be finely managed through processing, analyzing and predicting the historical energy consumption data, and the establishment of related energy management policies can be guided. Currently, accurate prediction of building energy consumption is critical for energy management and environmental protection. However, the traditional energy consumption prediction method has the problems of complex calculation, complicated data processing, low accuracy and the like, and is difficult to collect various environmental parameter data of the building. For this purpose, a corresponding technical solution needs to be designed to solve.
Disclosure of Invention
(One) solving the technical problems
Aiming at the defects of the prior art, the invention provides a rapid prediction method for building energy consumption, which solves the technical problems of complex calculation, complex data processing, low accuracy and the like of the traditional energy consumption prediction method, and is difficult to collect various environmental parameter data of a building.
(II) technical scheme
In order to achieve the above purpose, the invention is realized by the following technical scheme: a building energy consumption rapid prediction method comprises the following steps:
S1, acquiring environmental parameter data inside a building by deploying a sensor network, wherein the acquired data comprise real-time data or historical data;
s2, carrying out data preprocessing on the collected environmental parameter data, and providing a reliable data base for subsequent model training and prediction;
S3, training a building energy consumption prediction model by using the preprocessed environment parameter data and the corresponding energy consumption data and using a deep learning algorithm, and building a complex relation between the environment parameter and the energy consumption through model training;
S4, a trained deep learning model is utilized, environmental parameter data collected in real time are combined, rapid prediction of building energy consumption is carried out, the model outputs corresponding energy consumption prediction results, and the prediction process is carried out in a batch prediction or time-by-time prediction mode;
s5, generating an effective energy management strategy according to the energy consumption prediction result and the user demand, and formulating proper energy-saving measures according to the predicted energy consumption condition so as to optimize the equipment operation, thereby realizing the aims of high-efficiency utilization of energy and energy conservation and emission reduction.
Preferably, in step S1, the environmental parameter data inside the building includes temperature, humidity, illumination intensity, energy consumption, CO2 concentration, air flow rate, pressure, sound level, air quality index, personnel number activity, window status, equipment status, and external weather conditions.
Preferably, in step S2, the data preprocessing includes data cleaning, data conversion, data normalization and data set division;
The specific method for preprocessing the data comprises the following steps:
s201, cleaning the collected original data, removing noise, abnormal value and redundant information in the data, and using filtering, abnormal value detection and de-duplication technology and algorithm to ensure the data quality;
s202, converting the cleaned data, including feature extraction, dimension reduction and data reconstruction, wherein the converted data better reflects the features and modes of the data;
s203, carrying out normalization processing on the converted data, mapping the data to a specific range or distribution, wherein the normalization method comprises maximum and minimum normalization and normalization, and is used for ensuring that the data with different characteristics or attributes have similar scales and ranges;
s204, dividing the processed data set into a training set, a verification set and a test set, and using different data sets in subsequent model training and evaluation to verify the performance and generalization capability of the model.
Preferably, in step S3, a deep learning algorithm is used to train a building energy consumption prediction model, for establishing a complex relationship between environmental parameters and energy consumption, and the specific method steps include the following steps:
S301, deep convolution generates an countermeasure network: training a building energy consumption prediction model using DCGAN, DCGAN combines the ideas of convolving a neural network and generating an countermeasure network;
Generating energy consumption data from the environmental parameter data through a group of generator and discriminator networks, enabling the generated energy consumption data to be indistinguishable from real energy consumption data through countermeasure training, and capturing a complex nonlinear relation between the environmental parameter and the energy consumption;
S302, combination of variable self-encoder and generation of countermeasure network: training a building energy consumption prediction model using a method of combining VAE and GAN, the VAE being used to learn potential representations of environmental parameter data while the GAN is used to generate energy consumption data;
The change of the environmental parameters is controlled in the potential space, corresponding energy consumption data are generated, the complex relation between the environmental parameters and the energy consumption is learned, and the energy consumption is rapidly predicted;
S303, graph neural network: training a building energy consumption prediction model by using a GNN, wherein the GNN is a deep learning method suitable for processing graph structure data, the environmental parameter data in a building is represented as a graph structure, and the GNN learns the dependency relationship between environmental parameters and the complex relationship between energy consumption;
and information transmission and interaction are carried out between environmental parameter data at different positions inside the building through the model, so that quick prediction of energy consumption is realized.
Preferably, the deep convolution generates an countermeasure network:
loss function of generator network (G): min-log (D (G (z)));
Loss function of the arbiter network (D): min-log (D (x)) -log (1-D (G (z)));
combination of variable self-encoder and generation of countermeasure network:
Loss function of generator network (G): min-log (D (G (z))) + KL DIVERGENCE;
Loss function of the arbiter network (D): min-log (D (x)) -log (1-D (G (z)));
Graph neural network:
the propagation formula of the graph neural network: h_v { (l+1) } = f \left (\sum { u\in N (v) } \frac {1} { c } { u, v } } W { (l) } h_u { (l) } +w { (l) } h_v { (l) } right);
Wherein h_v { (l+1) } represents a hidden representation of node v at layer l+1, h_u { (l) } represents a hidden representation of node u at layer l, N (v) represents a set of neighbor nodes of node v, W { (l) } represents a weight matrix of layer l, c_ { u, v } represents an element in the normalized adjacency matrix;
the output layer calculation formula of the graph neural network: o_v=g_left (\sum_u\in N (v) } W { (L) } h_u { (L) } +w { (L) } h_v { (L) } right);
Where o_v represents the output of node v, L represents the last layer, W { (L) } represents the weight matrix of the last layer, and g represents the activation function of the output layer.
Min represents minimization, i.e. minimizing the objective function by adjusting the model parameters;
log represents a logarithmic function, typically using natural logarithms or 2-base logarithms, which is typically used in a loss function in deep learning to measure the difference between the predicted and the actual values of the model;
z represents random vectors in the potential space, also called noise vectors, and in generating the reactance network, the generator network receives as input a random potential vector z, which is mapped by learning to synthesized samples;
x represents input data, such as building environmental parameter data, where x may represent real-time or historical environmental parameter data in building energy consumption prediction;
KL DIVERGENCE denotes KL-divergence for measuring the difference between two probability distributions, in a variational self-encoder, KL-divergence is used to quantify the difference between the hidden variable distribution in the potential space and the standard normal distribution to ensure that the potential space has a certain continuity and regularity;
The diversity refers to the KL divergence, or to other functions that measure the difference between two probability distributions;
left and right are used to define the scope of an expression, separate it from other operators or expressions, for added readability and clarity;
sum represents a summation operator for summing a series of values, in a given expression, sum represents summing a specified value or expression;
frac represents a fraction, and the ratio of the numerator to the denominator representing a numerical value is used to represent the result of the division of the two expressions.
Preferably, in step S4, the specific method steps for performing the rapid prediction of building energy consumption include the following steps:
S401, based on generating a fast prediction of the antagonism network: training an energy consumption prediction model by using a generated countermeasure network structure, wherein a generator network receives environmental parameter data acquired in real time as input, generates a rapid prediction result of energy consumption by learning a complex relation between environmental parameters and energy consumption, and a discriminator network is used for evaluating the authenticity of the generated energy consumption prediction result, so that the generator can generate a high-quality energy consumption prediction result by performing countermeasure training on the generator and the discriminator;
S402, fast prediction based on attention mechanism: attention mechanisms are adopted to strengthen attention of the model to key features in the environmental parameters, and the attention mechanisms can automatically learn and select the environmental parameter features most relevant to energy consumption prediction by introducing attention layers into the model;
S403, fast prediction based on a graph neural network: constructing environmental parameter data in a building into a graph structure, learning the relation between nodes by using a graph neural network, predicting the energy consumption of the nodes, training the graph neural network, and considering the mutual influence in space;
s404, fast prediction based on transfer learning: and performing transfer learning in a new building environment by utilizing a pre-trained deep learning model, pre-training on a large-scale building environment data set, learning a general relation between environment parameters and energy consumption by the model, applying the pre-trained model to the new building environment, and rapidly predicting the energy consumption of the building through fine adjustment and adaptive training.
Preferably, in step S4, the specific method steps of the building energy consumption batch prediction or time-by-time prediction combined with time series analysis and deep learning include the following steps:
S4001, establishing a time sequence model: firstly, constructing preprocessed environment parameter data and corresponding energy consumption data into a time sequence data set, and modeling and analyzing the data by using an ARIMA or SARIMA time sequence analysis method to capture trend, periodicity and seasonal characteristics in the data;
S4002, a prediction model based on deep learning: modeling and predicting time sequence data by using a deep learning algorithm, such as a cyclic neural network or a long-short-term memory network, taking the time sequence data as input, learning a complex relation and a nonlinear mode in the data by using the model, and obtaining a prediction result of energy consumption by training the deep learning model;
S4003, batch prediction or time-by-time prediction: according to actual demands, a batch prediction or time-by-time prediction mode is selected to conduct energy consumption prediction, in batch prediction, environment parameter data in a period of time is used as input, a time sequence model and a deep learning model are used for simultaneously predicting energy consumption results in the period of time, in time-by-time prediction, environment parameter data of each time point is used as input, and energy consumption results of each time point are predicted through the model.
Preferably, in step S5, the generation of an effective energy management strategy combines intelligent optimization and feedback control methods, specifically by the following steps:
S501, establishing an optimization model: an optimization model is established by utilizing historical energy consumption data, environment parameter data and real-time energy consumption prediction results, the model aims at minimizing energy consumption and meeting user requirements and comfort requirements, and meanwhile, the optimization model is established and solved by using a mathematical programming method of linear programming, integer programming or dynamic programming in consideration of the operation limit of building equipment and the constraint of energy supply;
S502, dynamically adjusting strategies: based on the result of the optimization model, generating energy management strategies for different time periods, including on-off control of equipment, adjustment of temperature and humidity set values, coordinated operation of an illumination and air conditioning system, and dynamic adjustment according to actual requirements and environmental conditions, so as to achieve the aims of efficient utilization of energy, energy conservation and emission reduction;
S503, feedback control mechanism: introducing a real-time monitoring and feedback control mechanism to adjust and optimize the energy management strategy in real time, monitoring environmental parameters and energy consumption conditions in the building in real time through a sensor network and an intelligent control system, comparing and adjusting the environmental parameters and the energy consumption conditions with an optimization model according to monitoring results, and realizing real-time feedback control so that the energy management strategy is more suitable for actual conditions and changes;
S504, intelligent decision support: by combining artificial intelligence and machine learning technology, an intelligent decision support system is established, an energy management strategy is learned and optimized, and optimal energy management advice and decision support are automatically provided by analyzing a large amount of historical data and real-time monitoring data.
Preferably, in step S5, the making of the appropriate energy saving measure includes the following specific method steps:
S5001, intelligent equipment control: based on the energy consumption prediction result and the real-time environmental parameter data, adopting an intelligent control algorithm to dynamically control equipment in a building, and intelligently adjusting the running state, the working mode and the energy consumption level of the equipment according to the actual demand and the environmental conditions through a self-adaptive control strategy;
s5002, energy optimization scheduling: the operation time and the working sequence of each device in the building are reasonably arranged through an optimized scheduling algorithm so as to avoid energy consumption conflict and overlapping operation among the devices, and the optimized scheduling algorithm takes the energy efficiency, the operation cost and the user demand factors of the devices into consideration to schedule the operation of the devices in an optimal mode;
S5003, dynamic energy supply: the renewable energy source and the energy source storage technology are combined, the dynamic energy source supply and the flexible allocation of the energy source are realized, the renewable energy source is reasonably utilized according to the energy consumption prediction result and the real-time requirement, and the energy source storage equipment is used for flexibly allocating the energy source supply so as to meet the energy source requirement of the building.
Preferably, the specific method steps for optimizing the operation of the device comprise the following steps:
S5004, intelligent maintenance and fault early warning: monitoring the running state and performance parameters of the equipment by using a sensor network and a data analysis technology, analyzing the health condition of the equipment in real time, predicting the fault risk of the equipment, timely discovering the abnormality and the fault of the equipment by using an intelligent maintenance and fault early warning system, and taking corresponding maintenance measures;
S5005, optimizing the energy efficiency of the equipment: the energy efficiency performance of the equipment is improved through equipment improvement and optimization, advanced technical means such as an intelligent sensor, a high-efficiency motor and an energy-saving controller are utilized to upgrade and reform the equipment, so that the energy consumption is reduced, the energy efficiency level of the equipment is improved, and the equipment can achieve the optimal energy utilization effect in actual operation;
S5006, data-driven operation and maintenance decision: and (3) carrying out deep mining and analysis on the operation data of the building equipment by utilizing big data analysis and machine learning technology, and making reasonable equipment operation strategies and maintenance plans based on data-driven operation and maintenance decisions.
(III) beneficial effects
Compared with the prior art, the invention has the beneficial effects that: by collecting various environmental parameter data of the building and utilizing a deep learning algorithm to carry out model training and prediction, complex nonlinear relation can be processed, rapid and accurate prediction of building energy consumption is realized, the accuracy and stability of prediction are improved, and the method has high efficiency, flexibility and instantaneity and can provide powerful support for energy management, energy conservation and emission reduction; by combining big data analysis and utilizing a large amount of environmental parameter data and historical energy consumption data for model training, the law and trend of building energy consumption can be captured better; the real-time energy consumption prediction is provided, and the dynamic adjustment can be carried out according to the environmental parameter data collected in real time, so that the method is suitable for different use situations and environmental changes.
Drawings
FIG. 1 is a schematic diagram of the overall process steps of the present invention;
FIG. 2 is a schematic diagram of an environmental parameter data system within a building according to the present invention;
FIG. 3 is a schematic diagram of a data preprocessing system according to the present invention;
FIG. 4 is a schematic diagram illustrating steps of a data preprocessing method according to the present invention;
FIG. 5 is a schematic diagram of specific method steps for training a building energy consumption prediction model using a deep learning algorithm for establishing a complex relationship between environmental parameters and energy consumption in accordance with the present invention;
FIG. 6 is a schematic diagram of steps of a specific method of the present invention for rapid prediction of building energy consumption;
FIG. 7 is a schematic diagram of steps of a specific method of the present invention for batch prediction or time-by-time prediction of building energy consumption in combination with time series analysis and deep learning;
FIG. 8 is a schematic diagram of steps of the method of the present invention for generating an effective energy management strategy in combination with intelligent optimization and feedback control;
FIG. 9 is a schematic diagram showing the steps of the method for establishing suitable energy saving measures according to the present invention;
FIG. 10 is a schematic diagram of specific method steps for optimizing the operation of the apparatus of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1 to 10, an embodiment of the present invention provides a technical solution: a building energy consumption rapid prediction method comprises the following steps:
S1, acquiring environmental parameter data inside a building by deploying a sensor network, wherein the acquired data comprise real-time data or historical data;
s2, preprocessing the acquired environmental parameter data, wherein the preprocessing aims to improve the quality and accuracy of the data, and is used for providing a reliable data base for subsequent model training and prediction;
S3, training a building energy consumption prediction model by using the preprocessed environment parameter data and the corresponding energy consumption data and using a deep learning algorithm, and building a complex relation between the environment parameter and the energy consumption through model training to improve the accuracy and the stability of prediction;
S4, a trained deep learning model is utilized, environmental parameter data collected in real time are combined, rapid prediction of building energy consumption is carried out, the model outputs corresponding energy consumption prediction results, and the prediction process is carried out in a batch prediction or time-by-time prediction mode;
s5, generating an effective energy management strategy according to the energy consumption prediction result and the user demand, and formulating proper energy-saving measures according to the predicted energy consumption condition so as to optimize the equipment operation, thereby realizing the aims of high-efficiency utilization of energy and energy conservation and emission reduction.
Further more, in step S1, the environmental parameter data inside the building includes temperature, humidity, illumination intensity, energy consumption, CO2 concentration, air flow rate, pressure, sound level, air quality index, personnel number activity, window status, equipment status and external meteorological conditions, so as to increase accuracy and comprehensiveness of rapid prediction of building energy consumption;
The concentration of CO2 is an important index of the air quality in a building, is closely related to energy consumption, and a CO2 sensor is used for monitoring the concentration of CO2 in the building;
The air flow rate inside the building influences the consumption of energy sources, including the operation effect and the ventilation and heat dissipation effect of an air conditioning system, and an air speed sensor is used for measuring the air flow rate inside the building;
the pressure change of gas or liquid in the building is related to energy consumption, and the pressure change of a water supply system can influence the energy consumption of a water pump, so that the pressure condition in the building is monitored by using a pressure sensor;
noise levels in buildings are related to energy consumption, including increases in noise levels that can result in more frequent use of air conditioning systems to reduce discomfort, and sound sensors to measure noise levels in buildings;
Air quality inside a building is critical to the health and comfort of people, and an air quality sensor is used for monitoring AQI (advanced integrated circuits) in the building, wherein the AQI comprises particle concentration and volatile organic compound indexes;
the number of people and activity level in the building have important influence on energy consumption, and a people counter or a motion sensor is used for monitoring the number of people and activity condition in the building;
The status of the window (open, closed, tilted) plays an important role for ventilation, lighting and heat transfer inside the building, using a window status sensor to monitor the status of the window inside the building;
the states and the use conditions of various devices (air conditioners, illumination and elevators) in the building have direct influence on energy consumption, and the device state sensors are used for monitoring the on-off states and the energy consumption level information of the devices;
meteorological conditions (temperature, humidity, wind speed) around a building have an important impact on energy consumption, and meteorological sensors are used to acquire external meteorological data.
The data of various sensors are combined to obtain more accurate and comprehensive building environment parameters.
Further improved, in step S2, the data preprocessing includes data cleaning, data conversion, data normalization and data set division;
The specific method for preprocessing the data comprises the following steps:
s201, cleaning the collected original data, removing noise, abnormal value and redundant information in the data, and using filtering, abnormal value detection and de-duplication technology and algorithm to ensure the data quality;
s202, converting the cleaned data, including feature extraction, dimension reduction and data reconstruction, wherein the converted data better reflects the features and modes of the data;
s203, carrying out normalization processing on the converted data, mapping the data to a specific range or distribution, wherein the normalization method comprises maximum and minimum normalization and normalization, and is used for ensuring that the data with different characteristics or attributes have similar scales and ranges;
s204, dividing the processed data set into a training set, a verification set and a test set, and using different data sets in subsequent model training and evaluation to verify the performance and generalization capability of the model.
Comprehensively considering the steps of data acquisition, cleaning, conversion, normalization and the like, and providing a high-quality data base for subsequent data analysis and application;
according to the requirements of specific tasks, different data processing technologies and algorithms are flexibly selected and applied so as to improve the accuracy and usability of the data to the greatest extent;
The method can process various types of data sources, including sensor data, database data and log files, and is suitable for data processing requirements of different fields and application scenes.
Further improved, in step S3, a deep learning algorithm is used to train a building energy consumption prediction model for establishing a complex relationship between environmental parameters and energy consumption, and the specific method steps include the following steps:
s301, deep convolution generates an antagonism network (DCGAN): training a building energy consumption prediction model using DCGAN, DCGAN combines the ideas of Convolutional Neural Networks (CNNs) and generating countermeasure networks (GANs);
Generating energy consumption data from the environmental parameter data through a set of generator and discriminator networks, and enabling the generated energy consumption data to be indistinguishable from real energy consumption data through countermeasure training, so that a complex nonlinear relationship between the environmental parameter and the energy consumption can be captured;
S302, combination of variable self-encoder (VAE) and generation of a countermeasure network (GAN): training a building energy consumption prediction model using a method of combining VAE and GAN, the VAE being used to learn potential representations of environmental parameter data while the GAN is used to generate energy consumption data;
Through the combination, the change of the environmental parameters is controlled in the potential space, corresponding energy consumption data is generated, the complex relation between the environmental parameters and the energy consumption is learned, and the energy consumption can be rapidly predicted;
S303, graph Neural Network (GNN): training a building energy consumption prediction model by using a GNN, wherein the GNN is a deep learning method suitable for processing graph structure data, the environmental parameter data in a building is represented as a graph structure, and the GNN learns the dependency relationship between environmental parameters and the complex relationship between energy consumption;
and information transmission and interaction are carried out between environmental parameter data at different positions inside the building through the model, so that quick prediction of energy consumption is realized.
By combining the characteristics of the deep learning algorithm, the complex characteristics of the internal environment parameter data of the building are fully utilized, so that a more accurate and stable relation model between the environment parameters and the energy consumption is established.
Further improved, the deep convolution generates an antagonism network (DCGAN):
loss function of generator network (G): min-log (D (G (z)));
Loss function of the arbiter network (D): min-log (D (x)) -log (1-D (G (z)));
combination of variable self-encoder (VAE) and generation of a countermeasure network (GAN):
Loss function of generator network (G): min-log (D (G (z))) + KL DIVERGENCE;
Loss function of the arbiter network (D): min-log (D (x)) -log (1-D (G (z)));
Graph Neural Network (GNN):
the propagation formula of the graph neural network: h_v { (l+1) } = f \left (\sum { u\in N (v) } \frac {1} { c } { u, v } } W { (l) } h_u { (l) } +w { (l) } h_v { (l) } right);
Wherein h_v { (l+1) } represents a hidden representation of node v at layer l+1, h_u { (l) } represents a hidden representation of node u at layer l, N (v) represents a set of neighbor nodes of node v, W { (l) } represents a weight matrix of layer l, c_ { u, v } represents an element in the normalized adjacency matrix;
the output layer calculation formula of the graph neural network: o_v=g_left (\sum_u\in N (v) } W { (L) } h_u { (L) } +w { (L) } h_v { (L) } right);
Where o_v represents the output of node v, L represents the last layer, W { (L) } represents the weight matrix of the last layer, and g represents the activation function of the output layer.
Min represents minimization, i.e. minimizing the objective function by adjusting the model parameters;
log represents a logarithmic function, typically using either natural logarithms (e-based) or 2-based logarithms, which are typically used in loss functions in deep learning to measure the difference between the predicted and actual values of the model;
z represents random vectors in potential space, also called noise vectors, and in generating a reactance network (GAN), the generator network receives as input a random potential vector z, which is mapped by learning to synthesized samples;
x represents input data, such as building environmental parameter data, where x may represent real-time or historical environmental parameter data in building energy consumption prediction;
KL DIVERGENCE denotes KL-divergence for measuring the difference between two probability distributions, in a variational self-encoder (VAE) for quantifying the difference between a hidden variable distribution in a potential space and a standard normal distribution to ensure that the potential space has a certain continuity and regularity;
The diversity refers to the KL divergence, or to other functions that measure the difference between two probability distributions;
left and right are used to define the scope of an expression, separate it from other operators or expressions, for added readability and clarity;
sum represents a summation operator for summing a series of values, in a given expression, sum represents summing a specified value or expression;
frac represents a fraction, and the ratio of the numerator to the denominator representing a numerical value is used to represent the result of the division of the two expressions.
Further improved, in step S4, the specific method steps for performing the rapid prediction of building energy consumption include the following steps:
S401, based on generating a fast prediction of the antagonism network (GAN): training an energy consumption prediction model by using a generated countermeasure network (GAN) structure, wherein the generator network receives environmental parameter data acquired in real time as input, generates a fast prediction result of energy consumption by learning a complex relationship between environmental parameters and energy consumption, and the discriminator network is used for evaluating the authenticity of the generated energy consumption prediction result, and enables the generator to generate a high-quality energy consumption prediction result by performing countermeasure training on the generator and the discriminator;
S402, fast prediction based on attention mechanism: attention mechanisms are adopted to strengthen attention of the model to key features in environmental parameters, so that the accuracy and stability of energy consumption prediction are improved, the attention mechanisms can automatically learn and select environmental parameter features most relevant to energy consumption prediction by introducing attention layers into the model, and the model can more accurately capture important features affecting energy consumption, so that quick and accurate energy consumption prediction is realized;
S403, fast prediction based on Graph Neural Network (GNN): constructing environmental parameter data inside a building into a graph structure, learning the relationship between nodes by using a Graph Neural Network (GNN), and predicting the energy consumption of the nodes, wherein the GNN effectively processes the spatial relationship and interaction in the building, so that the complex relationship between the environmental parameters is better captured, the graph neural network is trained, the rapid prediction of the energy consumption of the building is realized, and the spatial interaction is considered;
s404, fast prediction based on transfer learning: and performing transfer learning in a new building environment by utilizing a pre-trained deep learning model, so as to realize quick and accurate energy consumption prediction, learning a general relation between environment parameters and energy consumption by the model through pre-training on a large-scale building environment data set, applying the pre-trained model to the new building environment, and quickly predicting the energy consumption of the building through fine adjustment and adaptive training.
Based on the generation of the countermeasure network (GAN), the attention mechanism, the Graph Neural Network (GNN) and the transfer learning, the rapid and accurate prediction of the building energy consumption is provided, and more efficient decision support is provided for the management of the building energy consumption, energy conservation and emission reduction.
Further improved, in step S4, the specific method steps of building energy consumption batch prediction or time-by-time prediction combined with time series analysis and deep learning include the following steps:
S4001, establishing a time sequence model: firstly, constructing preprocessed environment parameter data and corresponding energy consumption data into a time sequence data set, and modeling and analyzing the data by using an ARIMA (autoregressive moving average model) or SARIMA (seasonal autoregressive moving average model) time sequence analysis method to capture trend, periodicity and seasonal characteristics in the data;
S4002, a prediction model based on deep learning: modeling and predicting time series data by using a deep learning algorithm, such as a cyclic neural network (RNN) or a long-short-term memory network (LSTM), taking the time series data as input, learning a complex relation and a nonlinear mode in the data by using the model, and obtaining a prediction result of energy consumption by training the deep learning model;
S4003, batch prediction or time-by-time prediction: according to actual demands, a batch prediction or time-by-time prediction mode is selected to conduct energy consumption prediction, in batch prediction, environment parameter data in a period of time is used as input, a time sequence model and a deep learning model are used for simultaneously predicting energy consumption results in the period of time, in time-by-time prediction, environment parameter data of each time point is used as input, the energy consumption results of each time point are predicted through the model, and a prediction mode is flexibly selected according to actual demands.
By combining the advantages of time sequence analysis and deep learning, the rapid prediction of building energy consumption is realized by establishing a time sequence model and applying a deep learning algorithm, the flexibility of batch prediction or time-by-time prediction is provided, and the dynamic change and complex relation of energy consumption can be captured more accurately.
Further improved, in step S5, generating an effective energy management strategy combines intelligent optimization and feedback control methods, specifically by the following steps:
S501, establishing an optimization model: an optimization model is established by utilizing historical energy consumption data, environment parameter data and real-time energy consumption prediction results, the model aims at minimizing energy consumption and meeting user requirements and comfort requirements, and meanwhile, the optimization model is established and solved by using a mathematical programming method of linear programming, integer programming or dynamic programming in consideration of the operation limit of building equipment and the constraint of energy supply;
S502, dynamically adjusting strategies: based on the result of the optimization model, generating energy management strategies for different time periods, including on-off control of equipment, adjustment of temperature and humidity set values, coordinated operation of an illumination and air conditioning system, and dynamic adjustment according to actual requirements and environmental conditions, so as to achieve the aims of efficient utilization of energy, energy conservation and emission reduction;
S503, feedback control mechanism: introducing a real-time monitoring and feedback control mechanism to adjust and optimize the energy management strategy in real time, monitoring environmental parameters and energy consumption conditions in the building in real time through a sensor network and an intelligent control system, comparing and adjusting the environmental parameters and the energy consumption conditions with an optimization model according to monitoring results, and realizing real-time feedback control so that the energy management strategy is more suitable for actual conditions and changes;
S504, intelligent decision support: by combining artificial intelligence and machine learning technology, an intelligent decision support system is established, an energy management strategy is learned and optimized, and optimal energy management advice and decision support are automatically provided by analyzing a large amount of historical data and real-time monitoring data.
The advantages of intelligent optimization and feedback control are combined, and an efficient energy management strategy is realized by establishing an optimization model, a dynamic adjustment strategy, introducing a feedback control mechanism and an intelligent decision support system, and the strategies not only consider the targets of energy conservation and emission reduction, but also meet the requirements of users and comfort, and can be dynamically adjusted and optimized according to real-time conditions.
Further improved, in step S5, the making of suitable energy saving measures comprises the following specific method steps:
S5001, intelligent equipment control: based on the energy consumption prediction result and the real-time environmental parameter data, adopting an intelligent control algorithm to dynamically control equipment in a building, and intelligently adjusting the running state, the working mode and the energy consumption level of the equipment according to actual requirements and environmental conditions through a self-adaptive control strategy so as to reduce the energy consumption to the maximum extent;
S5002, energy optimization scheduling: the operation time and the working sequence of each device in the building are reasonably arranged through an optimized scheduling algorithm, so that energy consumption conflict and overlapping operation among the devices are avoided, the energy efficiency, the operation cost and user demand factors of the devices are considered by the optimized scheduling algorithm, the operation of the devices is scheduled in an optimal mode, and the aims of high-efficiency utilization of energy, energy conservation and emission reduction are achieved;
S5003, dynamic energy supply: the renewable energy source, such as solar energy and wind energy, and the energy storage equipment, including a battery and a heat storage device, are reasonably utilized according to the energy consumption prediction result and real-time requirements, so as to flexibly allocate the energy source to meet the energy source requirements of the building and reduce the dependence on the traditional energy source.
In a specific refinement, the specific method steps for optimizing the operation of the device comprise the following steps:
S5004, intelligent maintenance and fault early warning: monitoring the running state and performance parameters of the equipment by using a sensor network and a data analysis technology, analyzing the health condition of the equipment in real time, predicting the fault risk of the equipment, timely discovering the abnormality and the fault of the equipment by using an intelligent maintenance and fault early warning system, and taking corresponding maintenance measures to reduce the energy consumption of the equipment, improve the reliability and the service life of the equipment;
S5005, optimizing the energy efficiency of the equipment: the energy efficiency performance of the equipment is improved through equipment improvement and optimization, advanced technical means such as an intelligent sensor, a high-efficiency motor and an energy-saving controller are utilized to upgrade and reform the equipment, so that the energy consumption is reduced, the energy efficiency level of the equipment is improved, and the equipment can achieve the optimal energy utilization effect in actual operation;
S5006, data-driven operation and maintenance decision: the method comprises the steps of deep mining and analyzing operation data of building equipment by utilizing big data analysis and machine learning technology, extracting operation rules and optimization potential of the equipment, and making reasonable equipment operation strategies and maintenance plans based on data-driven operation and maintenance decisions so as to realize efficient operation of the equipment, reduce energy consumption and prolong equipment life.
By means of intelligent equipment control, energy optimization scheduling, dynamic energy supply, intelligent maintenance and fault early warning, equipment energy efficiency optimization, data-driven operation and maintenance decision and the like, the optimization of proper energy-saving measures and equipment operation is realized, and proper energy-saving measures and equipment operation optimization are formulated so as to achieve the aims of high-efficiency utilization of energy and energy conservation and emission reduction.
In summary, by collecting various environmental parameter data of a building and performing model training and prediction by using a deep learning algorithm, complex nonlinear relations can be processed, rapid and accurate prediction of building energy consumption is realized, the accuracy and stability of prediction are improved, and the method has high efficiency, flexibility and instantaneity and can provide powerful support for energy management, energy conservation and emission reduction; by combining big data analysis and utilizing a large amount of environmental parameter data and historical energy consumption data for model training, the law and trend of building energy consumption can be captured better; the real-time energy consumption prediction is provided, and the dynamic adjustment can be carried out according to the environmental parameter data collected in real time, so that the method is suitable for different use situations and environmental changes.
While the fundamental and principal features of the invention and advantages of the invention have been shown and described, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing exemplary embodiments, but may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present disclosure describes embodiments, not every embodiment is provided with a separate embodiment, and that this description is provided for clarity only, and that the disclosure is not limited to the embodiments described in detail below, and that the embodiments described in the examples may be combined as appropriate to form other embodiments that will be apparent to those skilled in the art.

Claims (10)

1. The method for quickly predicting the building energy consumption is characterized by comprising the following steps of:
S1, acquiring environmental parameter data inside a building by deploying a sensor network, wherein the acquired data comprise real-time data or historical data;
s2, carrying out data preprocessing on the collected environmental parameter data, and providing a reliable data base for subsequent model training and prediction;
S3, training a building energy consumption prediction model by using the preprocessed environment parameter data and the corresponding energy consumption data and using a deep learning algorithm, and building a complex relation between the environment parameter and the energy consumption through model training;
S4, a trained deep learning model is utilized, environmental parameter data collected in real time are combined, rapid prediction of building energy consumption is carried out, the model outputs corresponding energy consumption prediction results, and the prediction process is carried out in a batch prediction or time-by-time prediction mode;
s5, generating an effective energy management strategy according to the energy consumption prediction result and the user demand, and formulating proper energy-saving measures according to the predicted energy consumption condition so as to optimize the equipment operation, thereby realizing the aims of high-efficiency utilization of energy and energy conservation and emission reduction.
2. The method for quickly predicting the energy consumption of a building according to claim 1, wherein: in step S1, the environmental parameter data inside the building includes temperature, humidity, illumination intensity, energy consumption, CO2 concentration, air flow rate, pressure, sound level, air quality index, personnel number activity, window status, equipment status, and external weather conditions.
3. The method for quickly predicting the energy consumption of a building according to claim 1, wherein: in the step S2, the data preprocessing comprises data cleaning, data conversion, data normalization and data set division;
The specific method for preprocessing the data comprises the following steps:
s201, cleaning the collected original data, removing noise, abnormal value and redundant information in the data, and using filtering, abnormal value detection and de-duplication technology and algorithm to ensure the data quality;
s202, converting the cleaned data, including feature extraction, dimension reduction and data reconstruction, wherein the converted data better reflects the features and modes of the data;
s203, carrying out normalization processing on the converted data, mapping the data to a specific range or distribution, wherein the normalization method comprises maximum and minimum normalization and normalization, and is used for ensuring that the data with different characteristics or attributes have similar scales and ranges;
s204, dividing the processed data set into a training set, a verification set and a test set, and using different data sets in subsequent model training and evaluation to verify the performance and generalization capability of the model.
4. The method for quickly predicting the energy consumption of a building according to claim 1, wherein: in step S3, a deep learning algorithm is used to train a building energy consumption prediction model, which is used to build a complex relationship between environmental parameters and energy consumption, and the specific method steps include the following steps:
S301, deep convolution generates an countermeasure network: training a building energy consumption prediction model using DCGAN, DCGAN combines the ideas of convolving a neural network and generating an countermeasure network;
Generating energy consumption data from the environmental parameter data through a group of generator and discriminator networks, enabling the generated energy consumption data to be indistinguishable from real energy consumption data through countermeasure training, and capturing a complex nonlinear relation between the environmental parameter and the energy consumption;
S302, combination of variable self-encoder and generation of countermeasure network: training a building energy consumption prediction model using a method of combining VAE and GAN, the VAE being used to learn potential representations of environmental parameter data while the GAN is used to generate energy consumption data;
The change of the environmental parameters is controlled in the potential space, corresponding energy consumption data are generated, the complex relation between the environmental parameters and the energy consumption is learned, and the energy consumption is rapidly predicted;
S303, graph neural network: training a building energy consumption prediction model by using a GNN, wherein the GNN is a deep learning method suitable for processing graph structure data, the environmental parameter data in a building is represented as a graph structure, and the GNN learns the dependency relationship between environmental parameters and the complex relationship between energy consumption;
and information transmission and interaction are carried out between environmental parameter data at different positions inside the building through the model, so that quick prediction of energy consumption is realized.
5. The method for quickly predicting the energy consumption of a building according to claim 4, wherein: deep convolution generates an countermeasure network:
loss function of generator network (G): min-log (D (G (z)));
Loss function of the arbiter network (D): min-log (D (x)) -log (1-D (G (z)));
combination of variable self-encoder and generation of countermeasure network:
Loss function of generator network (G): min-log (D (G (z))) + KL DIVERGENCE;
Loss function of the arbiter network (D): min-log (D (x)) -log (1-D (G (z)));
Graph neural network:
the propagation formula of the graph neural network: h_v { (l+1) } = f \left (\sum { u\in N (v) } \frac {1} { c } { u, v } } W { (l) } h_u { (l) } +w { (l) } h_v { (l) } right);
Wherein h_v { (l+1) } represents a hidden representation of node v at layer l+1, h_u { (l) } represents a hidden representation of node u at layer l, N (v) represents a set of neighbor nodes of node v, W { (l) } represents a weight matrix of layer l, c_ { u, v } represents an element in the normalized adjacency matrix;
the output layer calculation formula of the graph neural network: o_v=g_left (\sum_u\in N (v) } W { (L) } h_u { (L) } +w { (L) } h_v { (L) } right);
Where o_v represents the output of node v, L represents the last layer, W { (L) } represents the weight matrix of the last layer, and g represents the activation function of the output layer.
Min represents minimization, i.e. minimizing the objective function by adjusting the model parameters;
log represents a logarithmic function, typically using natural logarithms or 2-base logarithms, which is typically used in a loss function in deep learning to measure the difference between the predicted and the actual values of the model;
z represents random vectors in the potential space, also called noise vectors, and in generating the reactance network, the generator network receives as input a random potential vector z, which is mapped by learning to synthesized samples;
x represents input data, such as building environmental parameter data, in building energy consumption prediction x represents real-time or historical environmental parameter data;
KL DIVERGENCE denotes KL-divergence for measuring the difference between two probability distributions, in a variational self-encoder, KL-divergence is used to quantify the difference between the hidden variable distribution in the potential space and the standard normal distribution to ensure that the potential space has a certain continuity and regularity;
The diversity refers to the KL divergence, or to other functions that measure the difference between two probability distributions;
left and right are used to define the scope of an expression, separate it from other operators or expressions, for added readability and clarity;
sum represents a summation operator for summing a series of values, in a given expression, sum represents summing a specified value or expression;
frac represents a fraction, and the ratio of the numerator to the denominator representing a numerical value is used to represent the result of the division of the two expressions.
6. The method for quickly predicting the energy consumption of a building according to claim 1, wherein: in step S4, the specific method steps for performing the rapid prediction of building energy consumption include the following steps:
S401, based on generating a fast prediction of the antagonism network: training an energy consumption prediction model by using a generated countermeasure network structure, wherein a generator network receives environmental parameter data acquired in real time as input, generates a rapid prediction result of energy consumption by learning a complex relation between environmental parameters and energy consumption, and a discriminator network is used for evaluating the authenticity of the generated energy consumption prediction result, so that the generator can generate a high-quality energy consumption prediction result by performing countermeasure training on the generator and the discriminator;
S402, fast prediction based on attention mechanism: attention mechanisms are adopted to strengthen attention of the model to key features in the environmental parameters, and the attention mechanisms automatically learn and select the environmental parameter features most relevant to energy consumption prediction by introducing attention layers in the model;
S403, fast prediction based on a graph neural network: constructing environmental parameter data in a building into a graph structure, learning the relation between nodes by using a graph neural network, predicting the energy consumption of the nodes, training the graph neural network, and considering the mutual influence in space;
s404, fast prediction based on transfer learning: and performing transfer learning in a new building environment by utilizing a pre-trained deep learning model, pre-training on a large-scale building environment data set, learning a general relation between environment parameters and energy consumption by the model, applying the pre-trained model to the new building environment, and rapidly predicting the energy consumption of the building through fine adjustment and adaptive training.
7. The method for quickly predicting the energy consumption of a building according to claim 6, wherein: in step S4, the specific method steps of the building energy consumption batch prediction or time-by-time prediction combined with time sequence analysis and deep learning include the following steps:
S4001, establishing a time sequence model: firstly, constructing preprocessed environment parameter data and corresponding energy consumption data into a time sequence data set, and modeling and analyzing the data by using an ARIMA or SARIMA time sequence analysis method to capture trend, periodicity and seasonal characteristics in the data;
S4002, a prediction model based on deep learning: modeling and predicting time sequence data by using a deep learning algorithm, such as a cyclic neural network or a long-short-term memory network, taking the time sequence data as input, learning a complex relation and a nonlinear mode in the data by using the model, and obtaining a prediction result of energy consumption by training the deep learning model;
S4003, batch prediction or time-by-time prediction: according to actual demands, a batch prediction or time-by-time prediction mode is selected to conduct energy consumption prediction, in batch prediction, environment parameter data in a period of time is used as input, a time sequence model and a deep learning model are used for simultaneously predicting energy consumption results in the period of time, in time-by-time prediction, environment parameter data of each time point is used as input, and energy consumption results of each time point are predicted through the model.
8. The method for quickly predicting the energy consumption of a building according to claim 1, wherein: in step S5, an effective energy management strategy is generated by combining an intelligent optimization and feedback control method, specifically through the following steps:
S501, establishing an optimization model: an optimization model is established by utilizing historical energy consumption data, environment parameter data and real-time energy consumption prediction results, the model aims at minimizing energy consumption and meeting user requirements and comfort requirements, and meanwhile, the optimization model is established and solved by using a mathematical programming method of linear programming, integer programming or dynamic programming in consideration of the operation limit of building equipment and the constraint of energy supply;
S502, dynamically adjusting strategies: based on the result of the optimization model, generating energy management strategies for different time periods, including on-off control of equipment, adjustment of temperature and humidity set values, coordinated operation of an illumination and air conditioning system, and dynamic adjustment according to actual requirements and environmental conditions, so as to achieve the aims of efficient utilization of energy, energy conservation and emission reduction;
S503, feedback control mechanism: introducing a real-time monitoring and feedback control mechanism to adjust and optimize the energy management strategy in real time, monitoring environmental parameters and energy consumption conditions in the building in real time through a sensor network and an intelligent control system, comparing and adjusting the environmental parameters and the energy consumption conditions with an optimization model according to monitoring results, and realizing real-time feedback control so that the energy management strategy is more suitable for actual conditions and changes;
S504, intelligent decision support: by combining artificial intelligence and machine learning technology, an intelligent decision support system is established, an energy management strategy is learned and optimized, and optimal energy management advice and decision support are automatically provided by analyzing a large amount of historical data and real-time monitoring data.
9. The method for quickly predicting the energy consumption of a building according to claim 1, wherein: in step S5, the making of the appropriate energy saving measures includes the following specific method steps:
S5001, intelligent equipment control: based on the energy consumption prediction result and the real-time environmental parameter data, adopting an intelligent control algorithm to dynamically control equipment in a building, and intelligently adjusting the running state, the working mode and the energy consumption level of the equipment according to the actual demand and the environmental conditions through a self-adaptive control strategy;
s5002, energy optimization scheduling: the operation time and the working sequence of each device in the building are reasonably arranged through an optimized scheduling algorithm so as to avoid energy consumption conflict and overlapping operation among the devices, and the optimized scheduling algorithm takes the energy efficiency, the operation cost and the user demand factors of the devices into consideration to schedule the operation of the devices in an optimal mode;
S5003, dynamic energy supply: the renewable energy source and the energy source storage technology are combined, the dynamic energy source supply and the flexible allocation of the energy source are realized, the renewable energy source is reasonably utilized according to the energy consumption prediction result and the real-time requirement, and the energy source storage equipment is used for flexibly allocating the energy source supply so as to meet the energy source requirement of the building.
10. The method for quickly predicting the energy consumption of a building according to claim 9, wherein: the specific method steps for optimizing the operation of the equipment comprise the following steps:
S5004, intelligent maintenance and fault early warning: monitoring the running state and performance parameters of the equipment by using a sensor network and a data analysis technology, analyzing the health condition of the equipment in real time, predicting the fault risk of the equipment, timely discovering the abnormality and the fault of the equipment by using an intelligent maintenance and fault early warning system, and taking corresponding maintenance measures;
S5005, optimizing the energy efficiency of the equipment: the energy efficiency performance of the equipment is improved through equipment improvement and optimization, advanced technical means such as an intelligent sensor, a high-efficiency motor and an energy-saving controller are utilized to upgrade and reform the equipment, so that the energy consumption is reduced, the energy efficiency level of the equipment is improved, and the equipment can achieve the optimal energy utilization effect in actual operation;
S5006, data-driven operation and maintenance decision: and (3) carrying out deep mining and analysis on the operation data of the building equipment by utilizing big data analysis and machine learning technology, and making reasonable equipment operation strategies and maintenance plans based on data-driven operation and maintenance decisions.
CN202410181691.3A 2024-02-18 2024-02-18 Rapid prediction method for building energy consumption Pending CN118114816A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202410181691.3A CN118114816A (en) 2024-02-18 2024-02-18 Rapid prediction method for building energy consumption

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202410181691.3A CN118114816A (en) 2024-02-18 2024-02-18 Rapid prediction method for building energy consumption

Publications (1)

Publication Number Publication Date
CN118114816A true CN118114816A (en) 2024-05-31

Family

ID=91211575

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202410181691.3A Pending CN118114816A (en) 2024-02-18 2024-02-18 Rapid prediction method for building energy consumption

Country Status (1)

Country Link
CN (1) CN118114816A (en)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118296569A (en) * 2024-06-05 2024-07-05 华中科技大学 Urban neighborhood energy consumption prediction method based on artificial intelligence algorithm
CN118311880A (en) * 2024-06-07 2024-07-09 天津阿尔法优联电气有限公司 Intelligent control method and system for driver controller based on data analysis
CN118333247A (en) * 2024-06-17 2024-07-12 远洋亿家物业服务股份有限公司 Property energy loss evaluation method and system based on deep learning
CN118393962A (en) * 2024-06-24 2024-07-26 江西江铃集团晶马汽车有限公司 Intelligent monitoring and energy efficiency improving method and system for energy consumption of passenger car production line
CN118393974A (en) * 2024-06-26 2024-07-26 厦门万安智能有限公司 Digital twinning-based hospital double-carbon monitoring method, system and storage medium
CN118534830A (en) * 2024-07-26 2024-08-23 富邦能源(山东)集团有限公司 Energy management system and method
CN118544552A (en) * 2024-07-30 2024-08-27 博创智能装备股份有限公司 Intelligent energy consumption management system of injection molding machine
CN118570021A (en) * 2024-07-31 2024-08-30 国网江西省电力有限公司南昌供电分公司 Intelligent control method and system for auxiliary equipment of transformer substation
CN118561118A (en) * 2024-08-01 2024-08-30 浙江新再灵科技股份有限公司 Elevator running state judging method, system, medium and electronic equipment
CN118393974B (en) * 2024-06-26 2024-10-15 厦门万安智能有限公司 Digital twinning-based hospital double-carbon monitoring method, system and storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111178626A (en) * 2019-12-30 2020-05-19 苏州科技大学 Building energy consumption prediction method and monitoring prediction system based on WGAN algorithm
CN114154689A (en) * 2021-11-18 2022-03-08 中南大学 Method and system for predicting single-machine energy consumption of cold water host and storage medium
CN117314094A (en) * 2023-10-07 2023-12-29 上海益邦智能技术股份有限公司 Energy management and control method and system based on AI intelligent park
US20240023028A1 (en) * 2022-11-22 2024-01-18 Intel Corporation Wireless network energy saving with graph neural networks

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111178626A (en) * 2019-12-30 2020-05-19 苏州科技大学 Building energy consumption prediction method and monitoring prediction system based on WGAN algorithm
CN114154689A (en) * 2021-11-18 2022-03-08 中南大学 Method and system for predicting single-machine energy consumption of cold water host and storage medium
US20240023028A1 (en) * 2022-11-22 2024-01-18 Intel Corporation Wireless network energy saving with graph neural networks
CN117314094A (en) * 2023-10-07 2023-12-29 上海益邦智能技术股份有限公司 Energy management and control method and system based on AI intelligent park

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118296569A (en) * 2024-06-05 2024-07-05 华中科技大学 Urban neighborhood energy consumption prediction method based on artificial intelligence algorithm
CN118296569B (en) * 2024-06-05 2024-08-13 华中科技大学 Urban neighborhood energy consumption prediction method based on artificial intelligence algorithm
CN118311880A (en) * 2024-06-07 2024-07-09 天津阿尔法优联电气有限公司 Intelligent control method and system for driver controller based on data analysis
CN118311880B (en) * 2024-06-07 2024-09-17 天津阿尔法优联电气有限公司 Intelligent control method and system for driver controller based on data analysis
CN118333247B (en) * 2024-06-17 2024-08-20 远洋亿家物业服务股份有限公司 Property energy loss evaluation method and system based on deep learning
CN118333247A (en) * 2024-06-17 2024-07-12 远洋亿家物业服务股份有限公司 Property energy loss evaluation method and system based on deep learning
CN118393962A (en) * 2024-06-24 2024-07-26 江西江铃集团晶马汽车有限公司 Intelligent monitoring and energy efficiency improving method and system for energy consumption of passenger car production line
CN118393974A (en) * 2024-06-26 2024-07-26 厦门万安智能有限公司 Digital twinning-based hospital double-carbon monitoring method, system and storage medium
CN118393974B (en) * 2024-06-26 2024-10-15 厦门万安智能有限公司 Digital twinning-based hospital double-carbon monitoring method, system and storage medium
CN118534830A (en) * 2024-07-26 2024-08-23 富邦能源(山东)集团有限公司 Energy management system and method
CN118544552A (en) * 2024-07-30 2024-08-27 博创智能装备股份有限公司 Intelligent energy consumption management system of injection molding machine
CN118544552B (en) * 2024-07-30 2024-10-18 博创智能装备股份有限公司 Intelligent energy consumption management system of injection molding machine
CN118570021A (en) * 2024-07-31 2024-08-30 国网江西省电力有限公司南昌供电分公司 Intelligent control method and system for auxiliary equipment of transformer substation
CN118561118A (en) * 2024-08-01 2024-08-30 浙江新再灵科技股份有限公司 Elevator running state judging method, system, medium and electronic equipment

Similar Documents

Publication Publication Date Title
CN118114816A (en) Rapid prediction method for building energy consumption
CN117713221B (en) Micro-inversion photovoltaic grid-connected optimization system
CN117439101B (en) Intelligent network for interaction of new energy and flexible load in power grid
CN116307652A (en) Artificial intelligent resource allocation method for intelligent power grid
Pandey et al. Applications of artificial intelligence in power system operation, control and planning: a review
CN116599151A (en) Source network storage safety management method based on multi-source data
CN118171788B (en) Collaborative optimization method for schedulable flexible resources and power grid planning layout
CN110880055A (en) Building intelligent ammeter system
CN118364887B (en) Intelligent scheduling method and system for energy management monitoring
CN117028147A (en) Wind power variable pitch control system and wind power variable pitch system
CN116227883A (en) Intelligent household energy management system prediction decision-making integrated scheduling method based on deep reinforcement learning
CN110033181B (en) Power generation equipment state evaluation method based on self-encoder
CN111815039A (en) Weekly scale wind power probability prediction method and system based on weather classification
CN117175695B (en) Photovoltaic micro-grid power generation method and system based on diesel generator set
CN117974395A (en) Environment-friendly community management method and system based on carbon emission monitoring
CN117767446A (en) Multi-source power scheduling method and system based on big data
CN113240217A (en) Photovoltaic power generation prediction method and device based on integrated prediction model
CN117833319A (en) Electric power peak regulation system based on Internet of things
CN117543835A (en) Intelligent power distribution control method and device for power distribution cabinet
CN117346129A (en) Temperature regulation and control system for low-temperature heating surface of high-temperature superheater of boiler
CN117217553A (en) Low-carbon operation method of cogeneration system
CN116826733A (en) Photovoltaic power prediction method and system
CN105652658B (en) Real power control system based on the analysis of photovoltaic DC field data and real time environment monitoring
CN111967689A (en) Model and method for wind power generation prediction by combining multivariate and stepwise linear regression and artificial neural network
Li et al. Optimal Deep Belief Networks for Energy Demand Forecasting Using a Developed Version of the Gorilla Troops Optimization Method

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