Building energy consumption control system based on deep learning and control method thereof

Abstract

The invention discloses a building energy consumption control system based on deep learning. The building energy consumption control system comprises a data acquisition module, a data processing module, a deep learning module, an optimization control module and a real-time control module, the invention further discloses a control method of the building energy consumption control system based on deep learning. The energy consumption prediction model is constructed through the long-short-term memory network, and the user behavior model is constructed through the convolutional neural network, so that the energy consumption optimization control strategy is generated through the reinforcement learning algorithm based on the output results of the energy consumption prediction model and the user behavior model, and the related data of the building energy consumption system is effectively utilized. Intelligent control over the building energy consumption system is achieved, changes of environmental parameters and user requirements are dynamically adapted by optimizing an energy consumption control strategy, and the comfort level of a user is improved while energy saving and consumption reduction are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of building energy consumption control, and specifically relates to a building energy consumption control system based on deep learning and a control method thereof. Background Art

[0002] With the increasingly serious global energy crisis and environmental pollution problems, building energy consumption control has become one of the important means to improve energy utilization efficiency, reduce energy consumption and reduce greenhouse gas emissions.

[0003] Traditional building energy control systems mainly rely on preset fixed strategies and rules to manage the building's heating, ventilation and air conditioning (HVAC) system, lighting system and other electrical equipment.

[0004] However, these systems are usually based on simple schedules or empirical adjustment strategies based on historical data, which makes it difficult to adapt to complex and changing building environments and user needs in real time. Summary of the invention

[0005] The purpose of the present invention is to provide a building energy consumption control system based on deep learning to solve the problems raised in the above background technology. The present invention provides a building energy consumption control system based on deep learning, which has the characteristics of effectively utilizing the relevant data of the building energy consumption system to realize intelligent control of the building energy consumption system, and dynamically adapting to changes in environmental parameters and user needs by optimizing the energy consumption control strategy, thereby achieving energy saving and consumption reduction while improving user comfort.

[0006] Another object of the present invention is to provide a control method for a building energy consumption control system based on deep learning.

[0007] To achieve the above-mentioned objectives, the present invention provides the following technical solutions: a building energy consumption control system based on deep learning, comprising a data acquisition module, a data processing module, a deep learning module, an optimization control module and a real-time control module, wherein; the data acquisition module is used to collect building environment data, building equipment status data and user behavior data in real time, and use the collected data and building historical data as the first data, and the building historical data includes building historical energy consumption data, building historical environment data, building historical equipment status data and user historical behavior data; the data processing module is used to preprocess and extract features of the first data to obtain first processed data; the deep learning module is used to construct an energy consumption prediction model and a user behavior model based on the first processed data; the optimization control module is used to generate an energy consumption optimization control strategy through a reinforcement learning algorithm based on the output results of the first processed data, the energy consumption prediction model and the user behavior model; the real-time control module is used to adjust the operating status of the building equipment according to the energy consumption optimization control strategy.

[0008] In the present invention, further, the deep learning module constructs an energy consumption prediction model through a long short-term memory network, and outputs the energy consumption prediction value for each future time period through the energy consumption prediction model; the deep learning module constructs a user behavior model through a convolutional neural network, and outputs the user behavior prediction results for each future time period through the user behavior model.

[0009] Further in the present invention, the reinforcement learning algorithm includes a deep Q-network algorithm.

[0010] Further in the present invention, the real-time control module is connected to and controlled by the building equipment through the intelligent controller, wherein the building equipment includes the HVAC system, lighting system and electrical equipment of the building.

[0011] Further in the present invention, building environment data includes but is not limited to temperature, humidity, light, air quality, noise level, wind speed, wind direction and air pressure.

[0012] Further in the present invention, the building equipment status data includes but is not limited to the HVAC system status, the lighting system status, the electrical equipment operation status, and the usage of elevators and escalators.

[0013] Further in the present invention, user behavior data includes but is not limited to room occupancy, user activities and device usage frequency.

[0014] The present invention further includes a monitoring and evaluation module, which is used to monitor system performance and evaluate energy consumption control effects.

[0015] In the present invention, the control method of a building energy consumption control system based on deep learning further comprises the following steps:

[0016] (1) collecting building environment data, building equipment status data and user behavior data in real time, and using the collected data and building history data as first data;

[0017] (2) preprocessing and feature extracting the first data to obtain first processed data;

[0018] (3) constructing an energy consumption prediction model and a user behavior model based on the first processed data through a deep learning algorithm;

[0019] (iv) generating an energy consumption optimization control strategy through a reinforcement learning algorithm based on the first processed data, the energy consumption prediction model, and the output result of the user behavior model;

[0020] (V) Adjust the operating status of building equipment according to the energy consumption optimization control strategy.

[0021] In step (iii), the method of constructing an energy consumption prediction model by a deep learning algorithm is: constructing an energy consumption prediction model by a long short-term memory network, and outputting energy consumption prediction values ​​for each future time period by the energy consumption prediction model;

[0022] In step (iii), the method of constructing a user behavior model through a deep learning algorithm is: constructing a user behavior model through a convolutional neural network, and outputting user behavior prediction results for future time periods through the user behavior model.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] 1. The present invention constructs an energy consumption prediction model through a long short-term memory network and a user behavior model through a convolutional neural network, thereby generating an energy consumption optimization control strategy through a reinforcement learning algorithm based on the output results of the energy consumption prediction model and the user behavior model, effectively utilizing the relevant data of the building energy consumption system, realizing intelligent control of the building energy consumption system, and dynamically adapting to changes in environmental parameters and user needs by optimizing the energy consumption control strategy, thereby achieving energy saving and consumption reduction while improving user comfort;

[0025] 2. The present invention utilizes the characteristics of LSTM network that is good at processing time series data. By constructing an energy consumption prediction model through LSTM network, the historical patterns and trends of building energy consumption can be effectively captured, thereby accurately predicting future energy consumption changes;

[0026] 3. The present invention combines environmental data, device status data and user behavior data to comprehensively consider the influence of multiple factors and improve prediction accuracy;

[0027] 4. The present invention constructs a user behavior model through CNN, which can more accurately predict the user's behavior pattern and needs, thereby optimizing energy consumption without affecting the user's comfort;

[0028] 5. According to the user's historical behavior data and real-time behavior data, the system can provide personalized energy consumption control solutions to further improve user experience and satisfaction;

[0029] 6. The energy consumption prediction model and the user behavior model of the present invention complement each other and together constitute an important part of the building energy consumption control system. The energy consumption prediction model provides macro energy consumption demand prediction, and the user behavior model provides micro behavior pattern prediction. The combination of the two can achieve more accurate and efficient energy consumption management;

[0030] 7. The optimization control module of the present invention can generate accurate energy consumption optimization control strategies based on the output of the energy consumption prediction model and the user behavior model, and realize intelligent management and optimization of building energy consumption;

[0031] 8. The present invention designs a user feedback mechanism and calculates indicators such as energy-saving effect, equipment operation efficiency and user satisfaction through an energy consumption control effect evaluation algorithm. Based on the evaluation results, the evaluation results are fed back to the deep learning module and the optimization control module to adjust the prediction model and control strategy to improve system performance.

[0032] 9. The present invention designs a real-time monitoring platform to display system performance, including energy consumption data, equipment status, user feedback, etc., and sets up a performance alarm mechanism. When the system performance does not meet the standards, an alarm notification is sent, emergency optimization measures are initiated, and periodic reports are generated to record the system operation status and optimization effects for reference by operation and maintenance personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a block diagram of the control system of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] Example 1

[0036] See also Figure 1 The present invention provides the following technical solutions: a building energy consumption control system based on deep learning, comprising a data acquisition module, a data processing module, a deep learning module, an optimization control module and a real-time control module, wherein; the data acquisition module is used to collect building environment data, building equipment status data and user behavior data in real time, and use the collected data and building history data as the first data, and the building history data includes building historical energy consumption data, building historical environment data, building historical equipment status data and user historical behavior data; the data processing module is used to preprocess and extract features of the first data to obtain first processed data; the deep learning module is used to construct an energy consumption prediction model and a user behavior model based on the first processed data; the optimization control module is used to generate an energy consumption optimization control strategy through a reinforcement learning algorithm based on the output results of the first processed data, the energy consumption prediction model and the user behavior model; the real-time control module is used to adjust the operating status of the building equipment according to the energy consumption optimization control strategy.

[0037] Specifically, the data acquisition module mainly collects data by installing sensors at key locations in the building, such as temperature sensors, humidity sensors, light sensors, occupancy sensors, electricity meters and water meters, etc., and collects various environmental and equipment data in real time through the Internet of Things technology, and transmits it to the data processing module for data processing. The data processing module can be implemented by being deployed in a central data processing center or a cloud platform.

[0038] Among them, building environment data includes but is not limited to temperature, humidity, light, air quality, noise level, wind speed, wind direction and air pressure, which can be collected through temperature sensors, humidity sensors, light sensors, air quality sensors, noise sensors, etc.

[0039] Building equipment status data includes but is not limited to HVAC system status (such as on / off status, set temperature, wind speed, etc.), lighting system status (such as on / off status, brightness level, etc.), electrical equipment operating status (such as on / off status, working mode, etc.) and the usage of elevators and escalators. It can be collected through smart meters, equipment sensors and smart appliances. Among them, equipment sensors are used to monitor the operating status of HVAC systems, lighting systems, etc. (such as switch status, operating mode, fault information), and smart appliances have networking functions and can provide their working status and energy consumption data in real time.

[0040] User behavior data includes but is not limited to room occupancy, user activities and device usage frequency, which can be collected through occupancy sensors, motion sensors and smart devices, where smart devices include but are not limited to smart phones and smart watches, to collect user activity data and usage patterns. This embodiment does not limit the specific data types and sensor types used for collection.

[0041] Example 2

[0042] The difference between this embodiment and Embodiment 1 is that: specifically, the deep learning module constructs an energy consumption prediction model through a long short-term memory network, and outputs energy consumption prediction values ​​for each future time period through the energy consumption prediction model.

[0043] Specifically, the data is processed uniformly through the data processing module, missing values ​​and outliers are cleaned, feature vectors are extracted from the cleaned and normalized data, and the building environment data, building equipment status data, user behavior data and building history data are combined to construct a multidimensional time series, wherein the building history data may include the relationship between the building history energy consumption data and the building history environment data, the building history equipment status data and the user history behavior data. After constructing the multidimensional time series, the multidimensional time series is divided into a training set, a validation set and a test set, and the LSTM layer is constructed and the time series features are extracted through the long short-term memory network LSTM algorithm, and the output of the LSTM layer is mapped to the predicted value through the fully connected layer, and the mean square error is used as the loss function and the Adam optimizer is used to optimize the model parameters, and the training is stopped when the loss on the validation set is no longer reduced to prevent overfitting.

[0044] Constructing the LSTM layer and extracting time series features specifically includes:

[0045] f t =σ(W f ·[h t-1 ,X t ]+b f );

[0046] i t =σ(W i ·[h t-1 ,X t ]+b i );

[0047] c t =f t ☉c t-1 +i t ⊙tanh(W c ·[h t-1 ,X t ]+b c );

[0048] o t =σ(W o ·[h t-1 ,X t ]+b o );

[0049] h t =o t ☉tanh(c t );

[0050] X t =[X env,t , X dev，t , X user,t , X his,t];

[0051] Where, t represents the time; f t is the output of the forget gate, indicating the gate vector that determines the specific information to be forgotten at the current moment; l t is the input gate output, indicating the gating vector that determines the update of specific information at the current moment; c t is the cell state, indicating the long-term memory of LSTM; h t is the hidden state, representing the short-term memory and output of LSTM; σ is the sigmoid activation function, used to compress the value between 0 and 1; tanh is the tanh activation function, used to compress the value between -1 and 1; W f , W i , W c and W o are all weight matrices; b f 、b i 、b c and b o are all bias vectors; X t is the input data at the current moment, X env，t is the building environment data, X dev，t is the building equipment status data, X user，t is user behavior data, X his，t For building history data.

[0052] Finally, the output h of the LSTM layer t Mapped to the predicted value through a fully connected layer.

[0053] By adopting the above technical solution and taking advantage of the LSTM network's ability to process time series data, an energy consumption prediction model is constructed through the LSTM network. This can effectively capture the historical patterns and trends of building energy consumption, thereby accurately predicting future energy consumption changes. By combining environmental data, equipment status data, and user behavior data, the impact of multiple factors can be comprehensively considered to improve prediction accuracy.

[0054] Example 3

[0055] The difference between this embodiment and Embodiment 1 is that: specifically, the deep learning module constructs a user behavior model through a convolutional neural network, and outputs user behavior prediction results for each future time period through the user behavior model.

[0056] Specifically, feature vectors are extracted from the cleaned and normalized user behavior data. The following is a user behavior model construction process of a preferred implementation of Example 1 of the present invention:

[0057] Process user behavior data into feature vector X user ;

[0058] Xuser =[X enter , X exit , X location , X device ];

[0059] Among them, X enter Indicates that the user enters the record, X exit Indicates that the user has exited the record, X location Indicates the user activity area, X device Indicates the user device usage.

[0060] Then, the user behavior feature vector is constructed into a time series according to the time step, and the time series of the user behavior data is input into the input layer of the convolutional neural network. If the user behavior model has i convolutional layers, the local features of the user behavior data are extracted as follows:

[0061] C i =ReLU(W conv，i *X user +b conv，i );

[0062] Among them, W conv，i represents the convolution kernel of the i-th convolutional layer, b conv，i Represents the bias of the i-th convolutional layer, and ReLU is the activation function.

[0063] After each convolutional layer, a pooling layer is added to reduce the size of the feature map, and the feature map output by the pooling layer is flattened and passed through a fully connected layer for user behavior prediction.

[0064] By adopting the above technical solutions and building a user behavior model through CNN, users' behavior patterns and needs can be predicted more accurately, thereby optimizing energy consumption without affecting user comfort. For example, the operating status of air conditioners, lighting and other equipment can be reasonably adjusted according to the user's entry and exit time and activity area. Based on the user's historical behavior data and real-time behavior data, the system can provide personalized energy consumption control solutions to further improve user experience and satisfaction.

[0065] It should be noted that, in the embodiment of the present invention, although both the energy consumption prediction model and the user behavior model are trained and constructed using user behavior data, there is no conflict between the two. The energy consumption prediction model focuses on the overall energy consumption of the building, including macro factors such as seasonal changes, daily usage patterns, and weather influences. The purpose is to predict the overall energy consumption demand of the building, and comprehensively analyze a variety of data (such as historical energy consumption, weather, equipment status, user behavior, etc.), focusing on macro energy consumption trends. The main purpose of the user behavior model is to analyze and predict the specific behaviors and activity patterns of users inside the building, such as room occupancy, time and frequency of using electrical equipment, etc. It focuses on the specific behaviors of individuals or groups inside the building, such as when someone enters the room, what equipment is used, the frequency and pattern of activities, and predicts the user's behavior in the future time period, that is, focusing on micro user behavior details.

[0066] In the embodiment of the present invention, the overall energy consumption management strategy can be formulated by referring to the prediction results of the energy consumption prediction model, such as adjusting the operation plan of the HVAC system, optimizing energy scheduling, etc., and the control strategy can be refined by referring to the prediction results of the user behavior model, such as dynamically adjusting the lighting and temperature according to the room occupancy, optimizing the equipment switching time according to the user's usage habits, etc. The two models complement each other and together constitute an important part of the building energy consumption control system. The energy consumption prediction model provides macro energy consumption demand prediction, and the user behavior model provides micro behavior pattern prediction. The combination of the two can achieve more accurate and efficient energy consumption management.

[0067] Example 4

[0068] The difference between this embodiment and embodiment 1 is that: Specifically, the specific method for the optimization control module to generate the energy consumption optimization control strategy through the deep Q network algorithm is:

[0069] (1) Set the state vector S t, S t is composed of the real-time data and forecast data at the current time t:

[0070]

[0071] in, represents the future energy consumption forecast value output by the energy consumption forecast model, Represents the future user behavior prediction results output by the user behavior model;

[0072] (2) Set the action vector a t, at may include but is not limited to adjusting temperature (increase / decrease), controlling lighting (on / off), adjusting ventilation mode, etc., which are not specifically limited in the embodiment of the present invention; setting action space A, for example, A = [adjust air conditioning temperature, turn on / off lighting, change ventilation mode], and the operator may set it specifically according to the actual situation;

[0073] (3) Set the reward function rt:

[0074] r t =W1·△E t +W2·G t +W3·M t ;

[0075] Among them, △E t G is the energy saving compared with the previous moment; t User comfort, which is specifically assessed by the comfort range set by the user; M t is equipment wear, indicating the impact of equipment operating status on equipment life; W1, W2, W3 weight coefficients;

[0076] (4) Set the Q network function to Q(s, a; θ) and the target Q network function to Q*(s, a; θ*), then the target Q value y t for:

[0077] y t =r t +γmax a Q * (s t+1 , a; θ * );

[0078] Target Q value y t is the current reward r t and the discounted sum of the maximum expected Q-value of the next state, r t To perform action a at the current time step t t The immediate reward obtained after the γ discount factor is used to balance the impact of current rewards and future rewards. The value range is [0, 1], Q * (s t+1 , a; θ * ) The target Q network at the next time step t+1 is s t+1 The Q value prediction for all possible actions a is as follows, θ* is the parameter of the target Q network, max a Q * (s t+1 , a; θ * ) is the Q value corresponding to the action that can produce the maximum Q value at the next time step t+1. The target Q value y t Used to calculate the loss function of the Q network to update the parameters of the Q network.

[0079] By adopting the above technical solutions, the optimization control module can generate accurate energy consumption optimization control strategies based on the outputs of the energy consumption prediction model and the user behavior model, and realize intelligent management and optimization of building energy consumption.

[0080] Example 5

[0081] This embodiment is different from Embodiment 1 in that: specifically, the real-time control module is connected to and controlled by the building equipment through the intelligent controller, wherein the building equipment includes the HVAC system, lighting system and electrical equipment of the building. The intelligent controller is provided with interfaces required for communicating with the building equipment, and communicates with the building equipment based on these interfaces to realize control of the building equipment.

[0082] Specifically, a control algorithm is developed in a real-time control module to implement the control algorithm on an intelligent controller. The algorithm receives an energy consumption optimization control strategy generated by an optimization control module and sends control instructions to corresponding devices through the intelligent controller.

[0083] Example 6

[0084] The difference between this embodiment and embodiment 5 is that: specifically, a multi-agent system (MAS) is combined with building energy consumption control. Specifically, different building equipment (such as HVAC system, lighting system, electrical equipment) are regarded as independent agents, and a central coordination agent is introduced to receive and integrate the information of all agents, and send the control information to each agent for execution to achieve global optimization through collaboration and improve overall energy efficiency.

[0085] In large buildings, different areas may have different environmental conditions and usage requirements. Each area is equipped with an intelligent agent to monitor and control the energy consumption of the area. The intelligent agents optimize the energy consumption management of the entire building through mutual communication and coordination. Different types of equipment (such as HVAC systems, lighting systems, electrical equipment, etc.) are controlled by different intelligent agents, and the intelligent agents achieve global optimization of energy consumption through collaboration. For example, when one area needs to increase cooling, other areas can appropriately reduce cooling needs to maintain overall energy consumption balance.

[0086] Example 7

[0087] The present embodiment is different from the first embodiment in that: specifically, it further includes a monitoring and evaluation module, which is used to monitor the system performance and evaluate the energy consumption control effect.

[0088] By adopting the above technical solutions, a user feedback mechanism can be designed, such as a user interface or mobile application, through which users can provide feedback on comfort and satisfaction. The energy consumption control effect evaluation algorithm is used to calculate indicators such as energy saving effect, equipment operation efficiency and user satisfaction. Based on the evaluation results, the deep learning module and the optimization control module are fed back to adjust the prediction model and control strategy to improve system performance.

[0089] In addition, a real-time monitoring platform can be developed to display system performance, including energy consumption data, equipment status, user feedback, etc. A performance alarm mechanism can be set up to send an alarm notification when system performance does not meet the standards, initiate emergency optimization measures, and generate periodic reports to record system operation status and optimization effects for reference by operation and maintenance personnel.

[0090] Example 8

[0091] Furthermore, the control method of a building energy consumption control system based on deep learning described in the present invention comprises the following steps:

[0092] (1) collecting building environment data, building equipment status data and user behavior data in real time, and using the collected data and building history data as first data;

[0093] (2) preprocessing and feature extracting the first data to obtain first processed data;

[0094] (3) constructing an energy consumption prediction model and a user behavior model based on the first processed data through a deep learning algorithm;

[0095] (iv) generating an energy consumption optimization control strategy through a reinforcement learning algorithm based on the first processed data, the energy consumption prediction model, and the output result of the user behavior model;

[0096] (V) Adjust the operating status of building equipment according to the energy consumption optimization control strategy.

[0097] In step (iii), the method of constructing an energy consumption prediction model by a deep learning algorithm is: constructing an energy consumption prediction model by a long short-term memory network, and outputting energy consumption prediction values ​​for each future time period by the energy consumption prediction model;

[0098] In step (iii), the method of constructing a user behavior model through a deep learning algorithm is: constructing a user behavior model through a convolutional neural network, and outputting user behavior prediction results for future time periods through the user behavior model.

[0099] In summary, the present invention constructs an energy consumption prediction model through a long short-term memory network, and constructs a user behavior model through a convolutional neural network, so as to generate an energy consumption optimization control strategy through a reinforcement learning algorithm based on the output results of the energy consumption prediction model and the user behavior model, effectively utilize the relevant data of the building energy consumption system, realize intelligent control of the building energy consumption system, and dynamically adapt to changes in environmental parameters and user needs by optimizing the energy consumption control strategy, so as to achieve energy saving and consumption reduction while improving user comfort; the present invention utilizes the characteristics of LSTM network that is good at processing time series data, and constructs an energy consumption prediction model through LSTM network, which can effectively capture the historical patterns and trends of building energy consumption, so as to accurately predict future energy consumption changes; the present invention combines environmental data, equipment status data and user behavior data to comprehensively consider the influence of multiple factors and improve prediction accuracy; the present invention constructs a user behavior model through CNN, which can more accurately predict the user's behavior pattern and needs, so as not to affect the user's comfort while optimizing energy consumption; the present invention can provide a personalized energy consumption control solution based on the user's historical behavior data and real-time behavior data, Further improve user experience and satisfaction; the energy consumption prediction model and user behavior model of the present invention complement each other and together constitute an important part of the building energy consumption control system. The energy consumption prediction model provides macro energy consumption demand prediction, and the user behavior model provides micro behavior pattern prediction. The combination of the two can achieve more accurate and efficient energy consumption management; the optimization control module of the present invention can generate accurate energy consumption optimization control strategies based on the outputs of the energy consumption prediction model and the user behavior model, and realize intelligent management and optimization of building energy consumption; the present invention designs a user feedback mechanism, calculates indicators such as energy-saving effects, equipment operating efficiency and user satisfaction through an energy consumption control effect evaluation algorithm, and based on the evaluation results, feeds back to the deep learning module and the optimization control module to adjust the prediction model and control strategy to improve system performance; the present invention designs a real-time monitoring platform, displays system performance through the real-time monitoring platform, including energy consumption data, equipment status, user feedback, etc., and sets a performance alarm mechanism. When the system performance does not meet the standard, an alarm notification is sent, and emergency optimization measures are initiated, and periodic reports are generated to record system operation status and optimization effects for reference by operation and maintenance personnel.

[0100] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A building energy consumption control system based on deep learning, characterized in that: It includes a data acquisition module, a data processing module, a deep learning module, an optimization control module and a real-time control module, among which; The data acquisition module is used to collect building environment data, building equipment status data and user behavior data in real time, and use the collected data and building history data as the first data, the building history data includes building history energy consumption data, building history environment data, building history equipment status data and user history behavior data; The data processing module is used to preprocess and extract features from the first data to obtain first processed data; The deep learning module is used to build an energy consumption prediction model and a user behavior model based on the first processed data; The optimization control module is used to generate an energy consumption optimization control strategy through a reinforcement learning algorithm based on the first processed data, the energy consumption prediction model and the output result of the user behavior model; The real-time control module is used to adjust the operating status of the building equipment according to the energy consumption optimization control strategy.

2. A building energy consumption control system based on deep learning according to claim 1, characterized in that: The deep learning module constructs an energy consumption prediction model through a long short-term memory network, and outputs energy consumption prediction values ​​for each future time period through the energy consumption prediction model; the deep learning module constructs a user behavior model through a convolutional neural network, and outputs user behavior prediction results for each future time period through the user behavior model.

3. The building energy consumption control system based on deep learning according to claim 1 is characterized in that: The reinforcement learning algorithm includes a deep Q-network algorithm.

4. The building energy consumption control system based on deep learning according to claim 1, characterized in that: The real-time control module is connected to and controlled by the building equipment through the intelligent controller, wherein the building equipment includes the building's HVAC system, lighting system and electrical equipment.

5. The building energy consumption control system based on deep learning according to claim 1, characterized in that: The building environment data includes, but is not limited to, temperature, humidity, light, air quality, noise level, wind speed, wind direction and air pressure.

6. The building energy consumption control system based on deep learning according to claim 1, characterized in that: The building equipment status data includes, but is not limited to, HVAC system status, lighting system status, electrical equipment operating status, and elevator and escalator usage.

7. The building energy consumption control system based on deep learning according to claim 1 is characterized by: The user behavior data includes but is not limited to room occupancy, user activities and device usage frequency.

8. The building energy consumption control system based on deep learning according to claim 1, characterized in that: It also includes a monitoring and evaluation module, which is used to monitor system performance and evaluate energy consumption control effects.

9. A control method for a building energy consumption control system based on deep learning according to any one of claims 1 to 8, characterized in that: The following steps are involved: (1) collecting building environment data, building equipment status data and user behavior data in real time, and using the collected data and building history data as first data; (2) preprocessing and feature extracting the first data to obtain first processed data; (3) constructing an energy consumption prediction model and a user behavior model based on the first processed data through a deep learning algorithm; (iv) generating an energy consumption optimization control strategy through a reinforcement learning algorithm based on the first processed data, the energy consumption prediction model, and the output result of the user behavior model; (V) Adjust the operating status of building equipment according to the energy consumption optimization control strategy.

10. The control method of a building energy consumption control system based on deep learning according to claim 9, characterized in that: In the step (iii), the method of constructing the energy consumption prediction model by the deep learning algorithm is: constructing the energy consumption prediction model by the long short-term memory network, and outputting the energy consumption prediction value of each future time period by the energy consumption prediction model; In the step (iii), the method of constructing a user behavior model through a deep learning algorithm is: constructing a user behavior model through a convolutional neural network, and outputting user behavior prediction results for each future time period through the user behavior model.

Images