Building energy-saving control method, system, equipment and medium

Abstract

The invention discloses a building energy-saving control method, system and device and a medium, and relates to the technical field of energy-saving control, and the method comprises the steps: constructing a self-organizing multi-agent, carrying out the prediction of a future power demand value through an LSTM deep learning model by a first-stage agent, and carrying out the prediction of a future power demand value through a second-stage agent; the optimal set temperature value and the optimal heating and refrigerating equipment power value are calculated according to the energy-saving target by using a quadratic programming QP method, and the secondary intelligent agent performs power optimization for different areas of the building by using a fuzzy logic algorithm. According to the method, building energy-saving control has global optimization and local fine adjustment through combination of the first-level intelligent agent and the second-level intelligent agent, building energy-saving control is more refined through the third-level intelligent agent, energy waste is reduced to the maximum extent while the comfort degree of each room is guaranteed, and the energy-saving effect is improved. The building energy-saving control is further refined to the room level from the first-level intelligent agent and the second-level intelligent agent, and personalized energy-saving optimization is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of energy-saving control, and particularly to a building energy-saving control method, system, device and medium. Background Art

[0002] Under the background of increasingly strict carbon emission supervision, the optimal control of building energy consumption has become an important research direction in the field of intelligent buildings. Traditional building energy-saving control methods mainly rely on schedule control, rule control based on set thresholds, or centralized energy consumption management systems to optimize the energy consumption of heating and ventilation systems; However, although the current deep learning-based load prediction methods can provide relatively accurate energy demand predictions, they usually adopt static models and are difficult to adapt to the dynamic changes of building energy consumption. Especially in multi-region and multi-room environments, they lack precise personalized regulation capabilities. Secondly, in terms of regional energy consumption optimization, existing fuzzy logic algorithms usually rely on fixed membership functions and rules and are difficult to cope with real-time environmental changes, resulting in insufficient flexibility of regulation strategies. In addition, building energy consumption anomaly detection in existing systems still relies on simple threshold determination or anomaly detection based on traditional statistical methods, lacking an efficient intelligent prediction mechanism, making it difficult to detect and handle abnormal energy consumption problems in a timely manner. Summary of the Invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides a building energy-saving control method to solve the problems that although the current deep learning-based load prediction methods can provide relatively accurate energy demand predictions, they usually adopt static models and are difficult to adapt to the dynamic changes of building energy consumption. Especially in multi-region and multi-room environments, they lack precise personalized regulation capabilities. Secondly, in terms of regional energy consumption optimization, existing fuzzy logic algorithms usually rely on fixed membership functions and rules and are difficult to cope with real-time environmental changes, resulting in insufficient flexibility of regulation strategies. In addition, building energy consumption anomaly detection in existing systems still relies on simple threshold determination or anomaly detection based on traditional statistical methods, lacking an efficient intelligent prediction mechanism, making it difficult to detect and handle abnormal energy consumption problems in a timely manner.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a building energy-saving control method, which includes: Constructing a self-organizing multi-agent, where the primary agent predicts the future power demand value through an LSTM deep learning model, calculates the optimal set temperature value and the optimal heating and cooling equipment power value according to the energy-saving target using the quadratic programming QP method, and the secondary agent uses a fuzzy logic algorithm to optimize the power for different regions of the building; The third-level agent processes the data collected from the rooms using the Variational Autoencoder (VAE), and performs short-term prediction of the collected data through the Adaptive Time Series Decomposition (ASTD) to optimize the power consumption of different rooms. Construct an intelligent adjacency matrix with an adaptive topology, calculate the interaction weights between agents, dynamically adjust the agent connection weights according to the optimization contribution degree of the agents, update the adjacency matrix and the optimization contribution degree, and optimize the communication frequency of the agents. Use the Q-learning algorithm to update the strategy, and calculate the similarity of the environmental state data for the third-level agent to optimize the strategy. Encrypt and store the device adjustment data, perform authentication, detect abnormal energy consumption, and issue early warnings for abnormalities.

[0006] As a preferred embodiment of the building energy-saving control method described in the present invention, wherein: the self-organizing multi-agent is constructed, and the first-level agent predicts the future power demand value through the LSTM deep learning model, and uses the Quadratic Programming (QP) method to calculate the optimal set temperature value and the optimal heating and cooling equipment power value according to the energy-saving target. The second-level agent uses the fuzzy logic algorithm to optimize the power consumption of different areas of the building, including The self-organizing multi-agent includes a first-level agent, a second-level agent, and a third-level agent; The first-level agent collects the real-time energy consumption data of the building, including the energy consumption of heating, lighting, and ventilation systems, and collects the total electricity consumption of the building through smart meters and current sensors. Adopt a pre-trained LSTM deep learning model to predict the future power load value of the building according to the historical power load value. Through the building thermodynamics model, predict the future power demand value of the heating and cooling equipment according to the specific heat capacity of air and the air mass flow rate. Use the Quadratic Programming (QP) method to establish an optimization objective to minimize the power demand value of the heating and cooling equipment. Set the temperature range and power range of the heating and cooling equipment as constraint conditions, and use a QP solver to calculate the optimal set temperature value and the optimal heating and cooling equipment power value. Based on camera devices, perform area detection through the YOLOv5 object detection algorithm and count the building area occupancy value. Calculate the comprehensive lighting power according to the area occupancy and the startup power of the lighting equipment. The second-level agent constructs a regional state model and uses the fuzzy logic algorithm for different areas of the building to calculate the temperature deviation value, humidity deviation value, and carbon dioxide concentration deviation value according to the difference between the optimal set temperature value, the historical appropriate humidity value, the historical appropriate carbon dioxide concentration value, and the average temperature detection value, average humidity detection value, and average carbon dioxide concentration value of different areas. The trapezoidal membership function is adopted to set the fuzzy rules as follows: if the temperature deviation value is high and the area occupancy value is high, then increase the power value of the heating and cooling equipment; if the humidity deviation value is high and the area occupancy value is high, then increase the power value of the heating and cooling equipment; if the temperature deviation value is low and the area occupancy value is low, then decrease the ventilation flow value; if the carbon dioxide concentration deviation value is high and the area occupancy value is high, then increase the ventilation flow value; The Mamdani fuzzy inference method is used to calculate the membership degree, and the centroid method is used to defuzzify, and the regulation coefficient is calculated; According to the regulation coefficient, the power value of the regional heating and cooling equipment and the regional ventilation flow value are adjusted respectively, and transmitted to the three-level intelligent agent.

[0007] As a preferred scheme of the building energy-saving control method described in the present invention, wherein: the three-level intelligent agent uses a variational autoencoder (VAE) to process the data collected from the rooms, and performs short-term prediction of the collected data through an adaptive time series decomposition (ASTD), and optimizes the power of different rooms, including, The three-level intelligent agent collects room data based on the building rooms within the building area through room sensor data, including room temperature data, room humidity data, and room carbon dioxide concentration data; The variational autoencoder (VAE) is used for data dimensionality reduction. Through the adaptive time series decomposition (ASTD), the time series data of the room-collected data is constructed, and the empirical mode decomposition (EMD) is used to decompose the room-collected data to obtain the IMF components and the residual term. Wavelet denoising is performed on the IMF components of the room-collected data, and the time series data of the collected data is reconstructed with the residual term; According to the denoised data of different collected data, a denoised time series is formed and a state space vector is formed. The Kalman filter is applied for short-term prediction, and the predicted values of the collected data, including room temperature data, room humidity data, and room carbon dioxide concentration data, are calculated respectively. The ventilation flow adjustment value of the room is determined according to the fuzzy logic algorithm, and the temperature compensation value is calculated according to the heat loss of the room due to ventilation. Combining with the building thermodynamics model, the final power value of the room heating and cooling equipment is determined.

[0008] As a preferred scheme of the building energy-saving control method described in the present invention, wherein: the construction of an adaptive topology intelligent adjacency matrix, calculating the interaction weights between the intelligent agents, and dynamically adjusting the intelligent agent connection weights according to the optimization contribution degree of the intelligent agents, and updating the adjacency matrix and the optimization contribution degree, optimizing the communication frequency of the intelligent agents, including, An adaptive topology is constructed based on three intelligent agents. According to the three intelligent agents in the building, the total number of secondary intelligent agents and the total number of tertiary intelligent agents are determined respectively. The adjacency matrix A is used to represent the interaction relationship between the intelligent agents. In the adjacency matrix A, if two intelligent agents are directly connected, it is represented as 1, otherwise it is represented as 0; Calculate the interaction weight between agents based on the temperature values of adjacent rooms; Calculate the optimization contribution degree of the agent according to the difference in the optimized energy consumption data of the agent, and dynamically adjust the connection weight of the agent. According to the historical threshold of the optimization contribution degree of the agent, distinguish the contribution of the agent to energy conservation, and dynamically optimize the agent topology according to the optimization contribution degree of the corresponding agent; Based on the sum of the mean and standard deviation of the agent connection weights as the weight threshold, if the agent connection weight is less than or equal to the weight threshold, disconnect the agent connection and update the adjacency matrix A; Recalculate the optimization contribution degree of the agent, calculate the interaction weight according to the ratio of the optimization contribution degree of a single agent to that of all agents, and calculate the optimized communication frequency value according to the product of the reference communication frequency of the agent and the interaction weight.

[0009] As a preferred solution of the building energy conservation control method described in the present invention, wherein: the Q-learning algorithm is used for policy update, and the similarity of the environmental state data of the three-level agent is calculated for policy optimization, including, Use the Q-learning algorithm to take the adjustment of heating / cooling equipment and ventilation control as the control decision of the agent, take the temperature, humidity, and energy consumption value as the environmental state, use the optimization contribution degree value as the calculation reward, update the decision based on the control decision and the environmental state, and adopt the ϵ-greedy strategy for exploration to update the Q-learning value table; For the three-level agent, calculate the difference in the environmental state data between adjacent agents as the similarity value, and determine the similarity threshold based on historical data. If the similarity value of each environmental state data is less than or equal to the similarity threshold of the corresponding environmental state data, it is determined that the agent environmental states are similar, and the Q-learning value table is shared to obtain the optimized agent control strategy.

[0010] As a preferred solution of the building energy conservation control method described in the present invention, wherein: the encrypted transmission and storage of the equipment adjustment data and the authentication pointer use the AES-256 encryption algorithm to encrypt and store the adjustment of the building heating and cooling equipment power value and the adjustment of the regional ventilation flow value according to the agent control strategy, use the TLS protocol for transmission encryption, use HMAC to calculate the authentication token for authentication, and generate a log file.

[0011] As a preferred solution of the building energy-saving control method described in the present invention, wherein: the energy consumption anomaly detection and anomaly warning are carried out by using a pre-trained LSTM-autoencoder to predict the energy consumption data of the heating, lighting, and ventilation systems, and anomaly detection is carried out according to the historical anomaly threshold. If the data prediction value is greater than or equal to the anomaly threshold, it is determined that the energy consumption is abnormal, and a warning signal is sent to the maintenance terminal, and the predicted data of the energy consumption anomaly is logged.

[0012] In a second aspect, the present invention provides a building energy-saving control system, including, including, An energy consumption data acquisition module that acquires the energy consumption data of the building's heating, lighting, and ventilation systems; A power demand prediction module that predicts the future power load demand through an LSTM deep learning model, calculates the future power demand of heating and cooling equipment, and uses the quadratic programming method to calculate the optimal set temperature value and the optimal heating and cooling equipment power value; A regional-level energy consumption optimization module that optimizes the power value of regional heating and cooling equipment and the ventilation flow value based on a fuzzy logic algorithm, calculates the deviation values of temperature, humidity, and CO 2 Concentration, sets fuzzy rules using trapezoidal membership functions, calculates membership degrees through the Mamdani fuzzy inference method, and uses the centroid method to defuzzify and calculate the regulation coefficient for energy consumption optimization; A short-term prediction module that uses adaptive time series decomposition to construct the time series of room acquisition data and performs short-term prediction. Short-term prediction of temperature, humidity, and carbon dioxide concentration is carried out through Kalman filtering, room-level ventilation control is optimized, the heat loss caused by ventilation is calculated, and temperature compensation calculation is carried out to finally determine the power value of room-level heating and cooling equipment; An agent topology optimization module that uses an adjacency matrix A to represent the interaction relationship between agents, dynamically adjusts the agent connection structure, calculates the change in temperature of adjacent rooms, determines the agent interaction weight, improves the accuracy of data sharing, and dynamically adjusts the agent connection weight based on the optimization contribution degree to optimize the agent network topology structure; A strategy optimization module that adopts the Q-learning reinforcement learning algorithm to enable agents to autonomously learn the optimal heating, cooling, and ventilation control strategies, calculates the environmental state similarity, and determines whether to share the Q-learning value table based on the similarity of temperature, humidity, and energy consumption data between adjacent agents; A data security module that uses AES-256 to encrypt and store device adjustment data and encrypts the communication between agents using the TLS protocol; An anomaly detection module that uses an LSTM-autoencoder for data prediction, determines whether to trigger a warning based on the historical anomaly threshold, and records the anomaly prediction data.

[0013] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the building energy-saving control method described in the first aspect of the present invention is implemented.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the building energy-saving control method described in the first aspect of the present invention is implemented.

[0015] The beneficial effects of the present invention are as follows: The combination of the first-level intelligent agent and the second-level intelligent agent enables global optimization and local fine-tuning of building energy-saving control. The third-level intelligent agent makes the building energy-saving control more refined. While ensuring the comfort of each room, it minimizes energy waste to the greatest extent, making the building energy-saving control further refined from the first-level intelligent agent and the second-level intelligent agent to the room level, realizing personalized energy-saving optimization, improving the energy utilization efficiency of the overall building management system, and through the dynamic adjustment strategy of the optimization contribution degree of the intelligent agent, the computational load of energy consumption optimization is more reasonably allocated. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0017] Figure 1 It is a schematic flow chart of the building energy-saving control method in Embodiment 1.

[0018] Figure 2 It is a schematic structural diagram of the building energy-saving control system in Embodiment 2. Detailed Embodiments

[0019] To make the above objects, features, and advantages of the present invention more obvious and understandable, the detailed embodiments of the present invention will be described in detail below with reference to the drawings in the specification.

[0020] Many specific details are set forth in the following description in order to fully understand the present invention, but the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar promotions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0021] Second, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an individual or alternative embodiment that is mutually exclusive with other embodiments.

[0022] Embodiment 1, referring to Figure 1 and Figure 2 , is the first embodiment of the present invention. This embodiment provides a building energy-saving control method, including the following steps: S1. Construct a self-organizing multi-agent system. Among them, the first-level agent predicts the future power demand value through an LSTM deep learning model, calculates the optimal set temperature value and the optimal heating and cooling equipment power value according to the energy-saving goal using the quadratic programming QP method, and the second-level agent uses a fuzzy logic algorithm to optimize the power for different areas of the building; Preferably, constructing a self-organizing multi-agent system, where the first-level agent predicts the future power demand value through an LSTM deep learning model, calculates the optimal set temperature value and the optimal heating and cooling equipment power value according to the energy-saving goal using the quadratic programming QP method, and the second-level agent uses a fuzzy logic algorithm to optimize the power for different areas of the building, including, The self-organizing multi-agent system includes a first-level agent, a second-level agent, and a third-level agent; The first-level agent collects the real-time energy consumption data of the building, including the energy consumption of heating, lighting, and ventilation systems, and collects the total electricity consumption of the building through smart meters (measuring the electricity consumption of each area) and current sensors (detecting the power load of equipment); Adopt a pre-trained LSTM deep learning model to predict the future power load value of the building according to the historical power load value; Through the building thermodynamics model, predict the future power demand value of the heating and cooling equipment according to the specific heat capacity of air and the air mass flow rate, expressed as: ; Where represents the power demand value of the heating and cooling equipment at the prediction time t, represents the specific heat capacity of air, represents the air flow rate, represents the set temperature, represents the room temperature at time t; Use the quadratic programming QP method to establish an optimization goal to minimize the power demand value of the heating and cooling equipment, expressed as: ; Where represents the optimization goal, T represents the total number of time, The power demand value of the heating and cooling equipment at time t denotes the human comfort index at time t (determined based on the Fanger thermal comfort model), the target PMV value (determined based on historical data); Set the temperature range and power range of the heating and cooling equipment as constraints, and use a QP solver to calculate the optimal set temperature value and the optimal power value of the heating and cooling equipment; Based on the camera device, perform area detection through the YOLOv5 object detection algorithm, and count the building area occupancy value; Calculate the comprehensive lighting power according to the area occupancy and the startup power of the lighting equipment; The secondary agent constructs a regional state model, and uses the fuzzy logic algorithm for different areas of the building. According to the difference between the optimal set temperature value, the historical suitable humidity value, and the historical suitable carbon dioxide concentration value and the average temperature detection value, the average humidity detection value, and the average carbon dioxide concentration value of different areas, calculate the temperature deviation value, the humidity deviation value, and the carbon dioxide concentration deviation value; Use the trapezoidal membership function to set the fuzzy rule as follows: if the temperature deviation value is high and the area occupancy value is high, then increase the power value of the heating and cooling equipment; if the humidity deviation value is high and the area occupancy value is high, then increase the power value of the heating and cooling equipment; if the temperature deviation value is low and the area occupancy value is low, then decrease the ventilation flow value; if the carbon dioxide concentration deviation value is high and the area occupancy value is high, then increase the ventilation flow value; Use the Mamdani fuzzy inference method to calculate the membership degree, and use the centroid method to defuzzify to calculate the regulation coefficient; Adjust the power value of the regional heating and cooling equipment and the regional ventilation flow value respectively according to the regulation coefficient, and transmit them to the tertiary agent.

[0023] Predict future power demand through LSTM, which can adjust the heating / cooling strategy in advance, reduce energy waste, use the QP method to calculate the optimal heating / cooling equipment power, ensure that the heating / cooling equipment meets the comfort requirements with the lowest energy consumption, avoid overloading of equipment during peak electricity consumption through power load forecasting, thereby reducing peak power demand and lowering the building operation cost. By using the thermal comfort model PMV to calculate the human comfort index, ensure that the adjustment of the heating / cooling system will not cause the temperature to be too low or too high, improve the user experience. By using the fuzzy logic algorithm to calculate the regional temperature, humidity, CO 2 deviation, make the heating / cooling power distribution of each area more reasonable, avoid unnecessary high-energy consumption operation, automatically calculate the optimal heating / cooling power and ventilation flow through Mamdani fuzzy inference without manual intervention, greatly reduce the management cost, and improve the system intelligence level. By using CO 2Concentration detection and ventilation control strategies are implemented to ensure that the carbon dioxide concentration does not exceed the standard, improve air quality, and reduce the discomfort of personnel caused by excessive CO 2 By automatically detecting personnel flow through cameras, the lighting power and the operation of HVAC equipment are dynamically adjusted to reduce the inaccuracy of manual settings; The combination of the first-level agent and the second-level agent endows the building energy-saving control with the dual capabilities of global optimization and local fine-tuning. The first-level agent provides prediction and control strategies for global building energy consumption optimization to ensure that the power demand of heating and cooling equipment is within a reasonable range. The second-level agent, based on the actual situation of specific areas, conducts fine-grained regulation of heating, ventilation, and lighting systems to improve the regional energy utilization efficiency. Through this two-layer optimization architecture, the energy management of the entire building can not only achieve overall energy savings at the macroscopic level but also perform dynamic adjustment for each area at the microscopic level to adapt to the actual needs of different times, occupancy situations, and environmental conditions. This method effectively reduces energy waste, lowers the overall building energy consumption, and at the same time ensures the comfort of the indoor environment, improving the efficiency and accuracy of building intelligent management.

[0024] S2, the third-level agent uses the variational autoencoder VAE to process the data collected from rooms and conducts short-term prediction of the collected data through the adaptive time series decomposition ASTD to optimize the power of different rooms; Preferably, the third-level agent uses the variational autoencoder VAE to process the data collected from rooms and conducts short-term prediction of the collected data through the adaptive time series decomposition ASTD to optimize the power of different rooms, including, Based on the building rooms within the building area, the third-level agent collects room data through room sensor data, including room temperature data, room humidity data, and room carbon dioxide concentration data; The variational autoencoder VAE is used for data dimensionality reduction. Through the adaptive time series decomposition ASTD, the time series data of the collected room data is constructed, and the empirical mode decomposition EMD is used to decompose the collected room data to obtain the IMF components and the residual term. Wavelet denoising is performed on the IMF components of the collected room data, and the time series data of the collected data is reconstructed with the residual term, expressed as: ; Where represents the denoised data at time t of the r-th type of collected data, m and k respectively represent the termination component index and the starting component index of the IMF components, represents the i-th component index of the r-th type of collected data, represents the residual term at time t of the r-th type of collected data; The denoised data of different collected data form a denoising time series and form a state space vector. Kalman filtering is applied for short-term prediction, and the predicted values of the collected data, including room temperature data, room humidity data, and room carbon dioxide concentration data, are calculated respectively. The room ventilation flow adjustment value is determined according to the fuzzy logic algorithm, and the temperature compensation value is calculated according to the heat loss of the room due to ventilation. Combining with the building thermodynamics model, the final power value of the room heating and cooling equipment is determined.

[0025] Through the agent's collection of sensor data at the room level, the refined monitoring of the building's internal environment can be realized. It can be independently optimized according to the specific environmental conditions of different rooms. By using the variational autoencoder (VAE) for data dimensionality reduction, the system can effectively reduce the redundant information of high-dimensional data, extract key features at the same time, improve the efficiency of data processing and the stability of prediction. Through the adaptive time series decomposition (ASTD), the three-level agent can construct time series data, making the collected room environmental variable data more predictable. In addition, the use of empirical mode decomposition (EMD) ensures the accuracy of short-term prediction. Through this series of data preprocessing methods, the three-level agent can effectively remove the random interference and abnormal fluctuations in the environmental monitoring data, making the subsequent prediction and control more accurate. Through the three-level agent, the building energy-saving control becomes more refined. While ensuring the comfort of each room, it minimizes energy waste to the greatest extent. Compared with the traditional fixed set temperature or the method based on regional average regulation, the three-level agent can perform personalized adjustment according to the real-time state of each room. It not only improves the energy utilization rate but also avoids the local overcooling or overheating problems that may be caused by regional-level control. The three-level agent provides an efficient and intelligent solution in data processing, prediction, and room-level energy consumption optimization, making the building energy-saving control further refined from the global level (first-level agent) and regional level (second-level agent) to the room level, realizing personalized energy-saving optimization and improving the energy utilization efficiency of the overall building management system.

[0026] S3. Construct an intelligent adjacency matrix with an adaptive topology, calculate the interaction weights between agents, dynamically adjust the agent connection weights according to the agent optimization contribution degree, update the adjacency matrix and the optimization contribution degree, and optimize the communication frequency of the agents. Preferably, constructing an intelligent adjacency matrix with an adaptive topology, calculating the interaction weights between agents, dynamically adjusting the agent connection weights according to the agent optimization contribution degree, updating the adjacency matrix and the optimization contribution degree, and optimizing the communication frequency of the agents includes An adaptive topology is constructed based on three agents. The total number of secondary agents and the total number of tertiary agents are determined according to the three agents in the building respectively. The adjacency matrix A is used to represent the interaction relationship between agents. In the adjacency matrix A, if two agents are directly connected, it is represented as 1, otherwise it is represented as 0; The interaction weight between agents is calculated based on the temperature values of adjacent rooms, which is expressed as: ; where represents the interaction weight between the i-th and j-th agents, represents the normalization parameter of temperature change, and represent the temperatures of the i-th and j-th adjacent rooms respectively; The optimization contribution degree of the agent is calculated based on the difference in optimized energy consumption data of the agent, and the connection weight of the agent is dynamically adjusted. According to the historical threshold of the optimization contribution degree of the agent, the contribution of the agent to energy conservation is distinguished, and the dynamic optimization of the agent topology is carried out according to the optimization contribution degree of the corresponding agent, which is expressed as: ; ; ; where represents the optimization contribution degree of the i-th agent, represents the energy consumption value before optimization, represents the energy consumption value after optimization, and represent the connection weights of the agent ij at time t and time t + 1 when the contribution degree is judged to be large respectively, and represent the connection weights of the agent ij at time t and time t + 1 when the contribution degree is judged to be small respectively, and represent the learning rates when the contribution degree is judged to be large and small respectively (set based on empirical data); Based on the sum of the mean and standard deviation of the agent connection weights as the weight threshold, if the agent connection weight is less than or equal to the weight threshold, the connection of the agent is disconnected, and the adjacency matrix A is updated; Recalculate the optimization contribution degree of the agent, calculate the interaction weight according to the ratio of the optimization contribution degree of a single agent to that of all agents, and calculate the optimized value of the communication frequency according to the product of the reference communication frequency of the agent and the interaction weight.

[0027] By constructing an intelligent adjacency matrix with an adaptive topology, the interaction structure of agents can be dynamically adjusted according to the change of the optimization contribution degree, enabling the system to adaptively optimize the energy consumption control strategy under different operating states, avoiding the inefficient interaction problems caused by a fixed topology. During the calculation process of the optimization contribution degree, agents can dynamically evaluate according to the energy consumption optimization effect, ensuring that resources are mainly allocated to the agents that can truly optimize the building energy consumption, reducing invalid calculations and data transmissions, thereby improving the computational efficiency and real-time response ability of the entire system. At the same time, since the connections of low-contribution agents will be disconnected when the weight is lower than the threshold, the system can automatically streamline the network structure, reduce redundant computational overhead, and make the building energy-saving control more efficient and flexible; The dynamic adjustment strategy of the optimization contribution degree of agents enables a more reasonable allocation of the computational load of energy consumption optimization. High-contribution agents undertake a larger proportion of computational tasks, while low-contribution agents reduce unnecessary calculations and data exchanges, thus avoiding waste of computational resources. In addition, by optimizing the communication frequency, the data exchange frequency of each agent is no longer fixed, but is proportional to its optimization contribution degree. High-contribution agents can exchange information with neighbors more frequently to ensure the effectiveness of the optimization strategy, while low-contribution agents reduce the communication load, further improving the computational efficiency and stability of the system.

[0028] S4. Use the Q-learning algorithm to update the strategy and optimize the strategy according to the similarity of the environmental state data of the three-level agents; Preferably, use the Q-learning algorithm to update the strategy and optimize the strategy according to the similarity of the environmental state data of the three-level agents, including, Use the Q-learning algorithm to take the adjustment of heating / cooling equipment and ventilation control as the control decisions of agents, take temperature, humidity, and energy consumption values as the environmental states, use the optimization contribution degree value as the calculation reward, update the decision based on the control decisions and environmental states, and adopt the ϵ-greedy strategy for exploration to update the Q-learning value table; For the three-level agents, calculate the difference between the environmental state data of adjacent agents as the similarity value, and determine the similarity threshold based on historical data. If the similarity value of each environmental state data is less than or equal to the similarity threshold of the corresponding environmental state data, it is judged that the environmental states of the agents are similar, and the Q-learning value table is shared to obtain an optimized agent control strategy.

[0029] By updating the policy through the Q-learning algorithm, the system enables the agent to autonomously learn the optimal heating, cooling, and ventilation control strategies without relying on fixed rules or manual intervention. Since Q-learning uses temperature, humidity, and energy consumption values as the environmental state and takes heating / cooling equipment adjustment and ventilation control as control decisions, the agent can continuously optimize the control strategy during long-term operation, enabling the system to adapt to changes in different building environments and improve the accuracy and adaptability of energy regulation. Using the optimization contribution as the calculation reward, the agent can find the best balance between energy conservation and comfort without simply pursuing minimizing energy consumption or maximizing comfort, thus avoiding problems such as excessive cooling or heating that may be caused by traditional energy-saving control strategies. For the three-level agent, agents with similar environmental states can share the Q-learning value table, improving the stability and reliability of the optimized control strategy. At the same time, since the sharing of the Q-learning value table is limited to agents with similar environmental states, it avoids incorrect learning caused by irrelevant environmental states and improves the effectiveness of the Q-learning learning process.

[0030] S5. Encrypt the transmission and storage of equipment adjustment data, perform authentication, detect energy consumption anomalies, and issue anomaly warnings. Preferably, encrypting the transmission and storage of equipment adjustment data and performing authentication means using the AES-256 encryption algorithm to encrypt and store the adjustment of the power value of the building heating and cooling equipment and the adjustment of the regional ventilation flow value according to the agent control strategy. Use the TLS protocol for transmission encryption, calculate the authentication token using HMAC for authentication, and generate a log file.

[0031] By encrypting the transmission and storage of equipment adjustment data, the system can effectively prevent unauthorized access and data tampering, ensuring the integrity and reliability of the agent control strategy. Since the adjustment of the power value of the heating and cooling equipment and the regional ventilation flow value directly affects building energy consumption optimization and comfort control, the security of the data is crucial. By using AES-256 encryption storage, even if the data is intercepted or leaked, unauthorized visitors cannot interpret it, thus preventing malicious tampering of equipment parameters from causing abnormal operation of the building system and ensuring the stability and security of the heating and cooling system.

[0032] Furthermore, detecting energy consumption anomalies and issuing anomaly warnings means using a pre-trained LSTM-autoencoder to predict the energy consumption data of the heating, lighting, and ventilation systems and performing anomaly detection based on historical anomaly thresholds. If the data prediction value is greater than or equal to the anomaly threshold, it is determined that there is an energy consumption anomaly, and a warning signal value is sent to the maintenance terminal, and the prediction data of the energy consumption anomaly is logged.

[0033] The energy consumption anomaly detection and warning system can effectively improve the safety and reliability of building energy consumption management, enabling the system to promptly detect and respond to abnormal energy consumption situations, avoiding potential losses caused by energy waste and abnormal equipment operation. By using a pre-trained LSTM-autoencoder for data prediction, the system can establish an accurate prediction model based on long-term energy consumption patterns, enhancing the accuracy of anomaly detection.

[0034] This embodiment also provides a building energy-saving control system, including an energy consumption data acquisition module that collects energy consumption data of the building's heating, lighting, and ventilation systems; a power demand prediction module that predicts future power load demands through an LSTM deep learning model, calculates the power demands of future heating and cooling equipment, and uses the quadratic programming method to calculate the optimal set temperature value and the optimal heating and cooling equipment power value; a regional-level energy consumption optimization module that optimizes the power values of regional heating and cooling equipment and ventilation flow rates based on the fuzzy logic algorithm, calculates the deviation values of temperature, humidity, and CO 2 concentration, sets fuzzy rules using trapezoidal membership functions, calculates membership degrees through the Mamdani fuzzy inference method, and uses the centroid method to defuzzify and calculate the regulation coefficients for energy consumption optimization; a short-term prediction module that uses adaptive time series decomposition to construct the time series of room-collected data and conducts short-term prediction, performs short-term prediction on temperature, humidity, and carbon dioxide concentration through Kalman filtering, optimizes room-level ventilation control, calculates the heat loss caused by ventilation, and conducts temperature compensation calculation to finally determine the power values of room-level heating and cooling equipment; an agent topology optimization module that uses the adjacency matrix A to represent the interaction relationships between agents, dynamically adjusts the agent connection structure, calculates the temperature changes in adjacent rooms, determines the agent interaction weights, improves the accuracy of data sharing, and dynamically adjusts the agent connection weights based on the optimization contribution degree to optimize the agent network topology structure; a strategy optimization module that adopts the Q-learning reinforcement learning algorithm to enable the agent to autonomously learn the optimal heating, cooling, and ventilation control strategies, calculates the environmental state similarity, and determines whether to share the Q-learning value table based on the similarity of temperature, humidity, and energy consumption data between adjacent agents; a data security module that uses AES-256 encryption to store device adjustment data and encrypts the communication between agents using the TLS protocol; an anomaly detection module that uses an LSTM-autoencoder for data prediction, determines whether to trigger an alarm based on historical anomaly thresholds, and records anomaly prediction data.

[0035] This embodiment also provides a computer device, which is applicable to the case of the building energy-saving control method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the building energy-saving control method proposed in the above embodiment.

[0036] The computer device may be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device may be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0037] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the building energy-saving control method proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read Only Memory (EPROM for short), Programmable Red-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0038] In summary, through the combination of the first-level agent and the second-level agent, the building energy-saving control in the present invention has global optimization and local refined adjustment. Through the third-level agent, the building energy-saving control becomes more refined. While ensuring the comfort of each room, it minimizes energy waste to the greatest extent, making the building energy-saving control further refined from the first-level agent and the second-level agent to the room level, realizing personalized energy-saving optimization, improving the energy utilization efficiency of the overall building management system, and through the dynamic adjustment strategy of the optimization contribution degree of the agent, the computational load of energy consumption optimization is more reasonably allocated.

[0039] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A building energy-saving control method, characterized in that: include: Construct a self-organizing multi-agent, where the first-level agent uses the LSTM deep learning model to predict future power demand values, uses the quadratic programming QP method to calculate the optimal set temperature value and the optimal heating and cooling equipment power value according to the energy-saving target, and the second-level agent uses the fuzzy logic algorithm to optimize the power for different areas of the building; The third-level agent uses variational autoencoder VAE to process room data collection, and uses adaptive time series decomposition ASTD to perform short-term prediction of collected data and optimize power for different rooms; Construct an intelligent adjacency matrix with adaptive topology, calculate the interaction weights between intelligent agents, dynamically adjust the connection weights of intelligent agents according to their optimization contributions, update the adjacency matrix and optimization contributions, and optimize the communication frequency of intelligent agents; Use the Q-learning algorithm to update the strategy and calculate the similarity of the environment state data for the three-level agents to optimize the strategy; The equipment adjustment data is encrypted for transmission and storage, and identity verification is performed to detect energy consumption anomalies and issue abnormal warnings.

2. The building energy saving control method according to claim 1, characterized in that: The self-organizing multi-agent is constructed, wherein the first-level agent predicts the future power demand value through the LSTM deep learning model, uses the quadratic programming QP method to calculate the optimal set temperature value and the optimal heating and cooling equipment power value according to the energy saving target, and the second-level agent uses the fuzzy logic algorithm to optimize the power for different areas of the building, including: The self-organizing multi-agent includes a primary agent, a secondary agent, and a tertiary agent; The first-level intelligent agent collects real-time energy consumption data of the building, including the energy consumption of heating, lighting, and ventilation systems, and collects the total electricity consumption of the building through smart meters and current sensors; Use a pre-trained LSTM deep learning model to predict the future power load value of the building based on the historical power load value; The future power demand of heating and cooling equipment is predicted based on the specific heat capacity of air and the air mass flow rate through the building thermodynamic model; The optimization objective is to minimize the power demand of heating and cooling equipment using the quadratic programming QP method; Set the temperature range and power range of the heating and cooling equipment as constraints, and use the QP solver to calculate the optimal set temperature value and the optimal heating and cooling equipment power value; Based on the camera device, the YOLOv5 target detection algorithm is used to detect the area and count the occupancy value of the building area; Calculate the comprehensive lighting power based on the area occupancy and lighting equipment startup power; The secondary agent constructs a regional state model and uses a fuzzy logic algorithm to calculate temperature deviation values, humidity deviation values, and carbon dioxide concentration deviation values ​​for different areas of the building according to the difference between the optimal set temperature value, the historical suitable humidity value, and the historical suitable carbon dioxide concentration value and the average temperature detection value, the average humidity detection value, and the average carbon dioxide concentration value of different areas; The fuzzy rules are set by using trapezoidal membership function: if the temperature deviation value is high and the area occupancy value is high, the power value of the heating and cooling equipment is increased; if the humidity deviation value is high and the area occupancy value is high, the power value of the heating and cooling equipment is increased; if the temperature deviation value is low and the area occupancy value is low, the ventilation flow value is reduced; if the carbon dioxide concentration deviation value is high and the area occupancy value is high, the ventilation flow value is increased; The Mamdani fuzzy reasoning method is used to calculate the membership degree, the centroid method is used to defuzzify, and the control coefficient is calculated; The power value of the regional heating and cooling equipment and the regional ventilation flow value are adjusted according to the control coefficient and transmitted to the third-level intelligent body.

3. The building energy saving control method according to claim 2, characterized in that: The three-level agent uses variational autoencoder VAE to process room data collection, and uses adaptive time series decomposition ASTD to perform short-term prediction of collected data, and optimizes power for different rooms, including: The third-level agent collects room data based on the building rooms in the building area through room sensor data, including room temperature data, room humidity data, and room carbon dioxide concentration data; Variational autoencoder (VAE) is used for data dimensionality reduction. Adaptive time series decomposition (ASTD) is used to construct the time series data of room data collection. Empirical mode decomposition (EMD) is used to decompose the room data collection to obtain IMF components and residual terms. Wavelet denoising is performed on the IMF components of the room data collection, and the time series data of the data collection is reconstructed with the residual terms. The denoised data of different collected data are used to form a denoised time series and a state space vector. Kalman filtering is used for short-term prediction, and the collected data, including the predicted values ​​of room temperature data, room humidity data, and room carbon dioxide concentration data, are calculated separately. The room ventilation flow adjustment value is determined based on the fuzzy logic algorithm, and the temperature compensation value is calculated based on the heat loss of the room due to ventilation. Combined with the building thermodynamics model, the final room heating and cooling equipment power value is determined.

4. The building energy-saving control method according to claim 3, characterized in that: The intelligent adjacency matrix of the adaptive topology is constructed, the interaction weights between intelligent agents are calculated, and the connection weights of intelligent agents are dynamically adjusted according to the optimization contribution of the intelligent agents, and the adjacency matrix and the optimization contribution are updated to optimize the communication frequency of the intelligent agents. include, An adaptive topology is constructed based on three intelligent agents. The total number of secondary intelligent agents and the total number of tertiary intelligent agents are determined according to the three intelligent agents in the building. The interaction relationship between the intelligent agents is represented by an adjacency matrix A. In the adjacency matrix A, if two intelligent agents are directly connected, it represents 1, otherwise it represents 0; Calculate the interaction weights between agents based on the temperature values ​​of adjacent rooms; Calculate the optimization contribution of the intelligent agent based on the energy consumption data difference of the intelligent agent optimization, and dynamically adjust the intelligent agent connection weight. According to the historical threshold of the intelligent agent optimization contribution, distinguish the contribution of the intelligent agent to energy saving, and dynamically optimize the intelligent agent topology according to the optimization contribution of the corresponding intelligent agent; Based on the sum of the mean and standard deviation of the agent connection weight as the weight threshold, if the agent connection weight is less than or equal to the weight threshold, the agent connection is disconnected and the adjacency matrix A is updated; Recalculate the optimization contribution of the intelligent agent, calculate the interaction weight based on the ratio of the optimization contribution of a single intelligent agent to all intelligent agents, and calculate the communication frequency optimization value based on the product of the intelligent agent's baseline communication frequency and the interaction weight.

5. The building energy-saving control method according to claim 4, characterized in that: The Q-learning algorithm is used to update the strategy, and the similarity of the three-level intelligent agent's environmental state data is calculated to optimize the strategy, including: Use the Q-learning algorithm to adjust the heating / cooling equipment and ventilation control as the control decision of the intelligent agent, and use the temperature, humidity, and energy consumption values ​​as the environmental state. The optimization contribution value is used as the calculation reward. The decision is updated based on the control decision and the environmental state. The ϵ-greedy strategy is used for exploration to update the Q-learning value table. For the three-level intelligent agents, the difference in environmental state data between adjacent intelligent agents is calculated as the similarity value, and the similarity threshold is determined based on historical data. If the similarity value of each environmental state data is less than or equal to the similarity threshold of the corresponding environmental state data, it is judged that the environmental states of the agents are similar, and the Q-learning value table is shared to obtain the optimized intelligent agent control strategy.

6. The building energy-saving control method according to claim 5, characterized in that: The encrypted transmission and storage of the equipment adjustment data and the identity authentication refer to the adjustment of the power value of the building heating and cooling equipment and the adjustment of the regional ventilation flow value by the intelligent control strategy using the AES-256 encryption algorithm for data encryption and storage, the TLS protocol for transmission encryption, the HMAC calculation of the identity authentication token, the identity authentication, and the generation of a log file.

7. The building energy-saving control method according to claim 6, characterized in that: The energy consumption anomaly detection and anomaly warning refers to using a pre-trained LSTM-autoencoder to predict the energy consumption data of the heating, lighting, and ventilation systems, and performing anomaly detection based on historical anomaly thresholds. If the data prediction value is greater than or equal to the anomaly threshold, it is judged as an energy consumption anomaly, and a warning signal is sent to the maintenance terminal, and the predicted data of the energy consumption anomaly is logged.

8. A building energy-saving control system, based on the building energy-saving control method according to any one of claims 1 to 7, characterized in that: include, Energy consumption data collection module, which collects energy consumption data of building heating, lighting and ventilation systems; The power demand prediction module predicts future power load demand through the LSTM deep learning model, calculates future heating and cooling equipment power demand, and uses the quadratic programming method to calculate the optimal set temperature value and the optimal heating and cooling equipment power value; The regional energy consumption optimization module optimizes the power value and ventilation flow value of regional heating and cooling equipment based on fuzzy logic algorithm, calculates the deviation value of temperature, humidity and CO2 concentration, and uses trapezoidal membership function to set fuzzy rules, calculates the membership degree through Mamdani fuzzy reasoning method, and uses the center of gravity method to solve the fuzzy calculation of the control coefficient to optimize energy consumption; The short-term prediction module uses adaptive time series decomposition to construct the time series of room collection data and conducts short-term predictions. It uses Kalman filtering to conduct short-term predictions of temperature, humidity, and carbon dioxide concentration, optimizes room-level ventilation control, calculates heat loss caused by ventilation, performs temperature compensation calculations, and ultimately determines the power value of room-level heating and cooling equipment; The agent topology optimization module uses the adjacency matrix A to represent the interaction relationship between agents, and dynamically adjusts the agent connection structure, calculates the temperature changes of adjacent rooms, determines the agent interaction weights, improves the accuracy of data sharing, dynamically adjusts the agent connection weights based on the optimization contribution, and optimizes the agent network topology structure; The strategy optimization module uses the Q-learning reinforcement learning algorithm to allow the intelligent agent to autonomously learn the optimal heating, cooling and ventilation control strategy, calculate the similarity of environmental states, and determine whether to share the Q-learning value table based on the similarity of temperature, humidity and energy consumption data between adjacent intelligent agents; Data security module, using AES-256 to encrypt storage devices to adjust data and TLS protocol to encrypt communications between agents; The anomaly detection module uses LSTM-autoencoder to predict data, determine whether to trigger an early warning based on historical anomaly thresholds, and record anomaly prediction data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the building energy-saving control method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the building energy-saving control method described in any one of claims 1 to 7 are implemented.

Images