Building energy management method based on big data

Abstract

The invention discloses a building energy management method based on big data, and relates to the technical field of building energy management, and the method comprises the steps: collecting building environment, room cold and hot demands and historical energy consumption data through a long and short term memory network LSTM in combination with Bayesian optimization, constructing a short-term and long-term cold load prediction model, and dynamically adjusting a cold and hot water supply strategy; the cold and hot water supply adjusting part enables cold and hot water conveying to accurately meet different room requirements on the basis of cold supply optimization parameters in combination with variable-frequency water pump regulation and control, water supply flow optimization of all floors and cold and hot water pipe network balance adjustment; an error feedback correction mechanism is introduced, the deviation between a predicted value and an actual load is monitored in real time, if the error exceeds a threshold value, an LSTM model is automatically adjusted, cooling parameters are updated, the prediction precision is improved, cooling and heating load regulation and control errors are reduced, and the influence of temperature fluctuation on user experience is reduced; aiming at different room temperature control requirements, personalized energy supply is carried out based on real-time temperature and humidity, air velocity and user preference.

Description

Technical Field

[0001] The present invention relates to the technical field of building energy management, and particularly to a building energy management method based on big data. Background Art

[0002] In current high-end hotels, such as five-star hotels, most of their central air-conditioning systems adopt a centralized cold and hot water supply method, and cold and hot water are transported to each floor and room through pipelines to maintain the indoor temperature. However, there are differences in the cold and hot demands of different rooms, and traditional building energy management methods are difficult to accurately match the actual loads of each room, resulting in unbalanced cold and hot water supply and affecting the experience of guests; it should be noted that the unbalanced cold and hot water supply here does not refer to the adjustment by users using the remote control in the room, but refers to the problems existing in the cold and hot water supply level of the central air-conditioning system.

[0003] Traditional central air-conditioning systems mostly adopt a control method with fixed water supply temperature and flow rate, and rely on the feedback of return water temperature for adjustment. However, this method can only reflect the overall load situation and cannot sense the temperature demands of individual rooms, resulting in excessive cooling in some rooms and insufficient cooling in some rooms; in addition, the changes in cold and heat loads are dynamic. In the case of large fluctuations in occupancy rate or significant day-night temperature differences, the cold and hot water demands will change violently in a short period of time, and traditional adjustment methods are difficult to respond in a timely manner, easily leading to large temperature fluctuations in some rooms.

[0004] Generally speaking, the traditional building energy management mode lacks the ability of dynamic adjustment based on real-time data. In large buildings, such as high-end places like five-star hotels, it is difficult to meet the requirements of cold and hot water supply, affecting the comfort of guests and the energy use efficiency; therefore, how to optimize the cold and hot water supply so that it can be dynamically adjusted according to the floor, room status and occupancy situation has become an important research direction for optimizing the energy management of large buildings such as hotels. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a building energy management method based on big data to solve the problems that the cooling method of the traditional central air-conditioning system is fixed, it is difficult to match the cold and hot water supply with the demands of different rooms, there is insufficient cooling on high floors, overcooling on low floors, and serious energy waste in empty rooms.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: An embodiment of the present invention provides a building energy management method based on big data, which includes, Step S1, collecting building environment data and room cold and hot demand data, and monitoring operating parameters; Step S2: Based on the data collected in Step S1, calculate the predicted values of cooling and heating loads and generate cooling optimization parameters. Step S3: According to the cooling optimization parameters generated in Step S2, adjust the supply of chilled and hot water and calculate the flow regulation data. In Step S2, the adjustment of the supply of chilled and hot water includes dynamic adjustment of the water supply pressure of the variable-frequency water pump, optimization of the water supply flow rate on each floor, and dynamic balance control of the chilled and hot water pipe networks. The flow regulation data includes the set values of the water supply flow rate for each room and the set values of the water supply pressure of the water pumps on different floors. Step S4: Based on the flow regulation data calculated in Step S3, perform room temperature control and calculate the room temperature control strategy. Step S5: According to the room temperature control strategy calculated in Step S4, link the occupancy status data and calculate the final air-conditioning operation mode. Step S6: According to the air-conditioning operation mode calculated in Step S5, perform the regulation of the central air-conditioning system, make the delivery of chilled and hot water match the needs of each room, and conduct the distribution of cooling and heating loads.

[0008] As a preferred solution of the building energy management method based on big data according to the present invention, wherein: the building environment data includes weather change information, room orientation information, and external temperature and humidity data. The room cooling and heating demand data includes the room occupancy rate, the historical temperature preferences of the occupants, and the energy consumption of the internal equipment in the room. The operating parameters include the current cooling capacity of the central air-conditioning system, the chilled water supply and return water pressures, and the room temperature set values.

[0009] As a preferred solution of the building energy management method based on big data according to the present invention, wherein: the predicted values of cooling and heating loads are used to predict the future cooling load demand through the long short-term memory network LSTM and the Bayesian optimization algorithm, and the cooling strategy of the central air-conditioning system is adjusted. The cooling optimization parameters include the predicted cooling load demand for each room, the adjustment amount of the chilled and hot water supply, and the target water supply pressure of the variable-frequency water pump.

[0010] As a preferred solution of the building energy management method based on big data according to the present invention, wherein: in Step S2, the method of predicting the future cooling load demand through the long short-term memory network LSTM and the Bayesian optimization algorithm includes: Collect the data of the past time periods as training data, including environmental variables, historical loads, occupancy conditions, etc. The time period is set to 10 - 20 minutes, and the total time span is set to 24 hours or 48 hours. The data matrix is expressed as: , Among them, is the training data matrix, is the feature vector at the -th time step, is the time step index, is the total number of time steps of the training data; Perform model training of LSTM. LSTM adopts time series modeling, and the input is historical data , and the output is the future cooling load demand , and the calculation process includes: , , Among them, is the LSTM hidden state, is the hidden state at the previous time step, is the hidden state weight matrix, is the input weight matrix, is the input data, is the hidden state bias, is the activation function, and the hyperbolic tangent function is adopted here, is the predicted cooling load, is the output weight matrix, is the output bias; Adopt Gaussian process regression GPR for Bayesian optimization, and the goal is to minimize the prediction error. The optimization process is expressed as: , Among them, is the loss function, is the cooling load predicted by the model, is the actual cooling load, Bayesian optimization calculates the optimal value of the hyperparameter , and the calculation formula is: , Among them, is the set of hyperparameters to be optimized, represents the expectation of the loss function, is the optimized hyperparameter.

[0011] As a preferred solution of the building energy management method based on big data described in the present invention, among them: in step S2, in the step of adjusting the cooling strategy of the central air conditioning system, calculate the cooling optimization parameters, including: The room cooling load demand, and the calculation formula is: , Among them, is the room cooling load demand, is an empirical regression function, and multiple linear regression is used for modeling, is the number of occupants, is the indoor temperature, is the outdoor temperature, is the relative air humidity, The adjustment amount of cold and hot water supply, and the calculation formula is: , where, is the adjustment amount of cold and hot water supply, is an empirical function, and logistic regression is used for modeling, is the current water supply pressure, is the water supply temperature, The target water supply pressure of the water pump, and the calculation formula is: , where, is the target water supply pressure, is the pressure prediction function, and support vector regression SVR is used for calculation, is the floor height factor; Introduce a dynamic error correction mechanism: , If , the LSTM model will be automatically adjusted and the cooling parameters will be re-optimized, where, is the relative error, is the predicted cooling load, is the actual cooling load, is the error threshold, set to 2% - 5%.

[0012] As a preferred solution of the building energy management method based on big data described in the present invention, where: in step S3, the methods of dynamically adjusting the water supply pressure of the variable-frequency water pump, optimizing the water supply flow of each floor, and dynamically balancing the control of the cold and hot water pipe network include: Set the target water supply pressure , and calculate the water pump power to match the cooling load demand of each floor. The formula is: , where, is the water pump power, is the current water supply flow, is the target water supply pressure, is the water pump system characteristic coefficient, calibrated based on the pump efficiency curve; Optimize the water supply flow of each floor according to the occupancy rate, cold and heat load demands, and historical data of each floor , the optimization formula is: , Among them, is the water supply flow rate for the floor, is predicted using a multiple linear regression model, is the current number of occupied rooms on the floor, is the floor height, is the outdoor temperature; Dynamically balance the cold and hot water pipe networks, and the cold and hot water supply is dynamically adjusted based on the temperature difference between supply and return water and flow feedback data. The adjustment formula is: , Among them, is the water supply pressure adjustment amount. Here, is adjusted using PID control, is the return water temperature, is the return water flow rate; Finally, determine the time step for the dynamic adjustment of the cold and hot water supply. The determination formula is: , Among them, is the water supply adjustment time interval, uses an adaptive time step algorithm, is the number of occupied rooms on the current floor, is the outdoor temperature, is the pressure adjustment amount.

[0013] As a preferred solution of the building energy management method based on big data described in the present invention, among them: the room temperature control includes real-time monitoring of the room temperature and humidity status, air velocity analysis, and identification of the temperature preferences of the occupants; The room temperature control strategy includes increasing the cooling capacity for sunny rooms during the high-temperature period at noon, reducing the cooling capacity for shady rooms, and adjusting the target temperature in combination with the personalized settings of the occupants.

[0014] As a preferred solution of the building energy management method based on big data described in the present invention, among them: the occupancy status data includes room occupancy, reservation system status, access control data, and room equipment usage records; The air conditioner operation mode includes maintaining low-power operation at 26°C - 28°C for unoccupied rooms, starting to cool down 10 minutes - 20 minutes before guests check in, reducing the cooling supply when leaving the room for a short time, and restoring the set temperature when returning.

[0015] As a preferred solution of the building energy management method based on big data described in the present invention, among them: in step S5, the steps of linking the occupancy status data and calculating the final air conditioner operation mode are, Collect reservation system, access control data and room equipment usage records, analyze occupancy status, and construct a data matrix, which can be expressed as: , in, is the occupancy status dataset, For the The occupancy status data of time steps, is the time span of the occupancy data, is the time step index; Occupancy status data includes: Booking information: booked check-in time and early check-out, Access control data: time series of entering and leaving the room, Equipment use: usage of indoor facilities such as air conditioning and lighting; Long short-term memory network LSTM and logistic regression LR are used to identify user patterns. The identification formula is: , in, is the probability of users leaving for a short time, is the sigmoid activation function, is the pattern recognition weight matrix, For the check-in dataset, is the bias term.

[0016] As a preferred solution of the big data-based building energy management method described in the present invention, in step S5, the step of linking the occupancy status data and calculating the final air conditioning operation mode also includes: Before your scheduled check-in Start the air conditioner in minutes and calculate the advance time: , in, Start the air conditioner in advance. Using a linear regression model, is the outdoor temperature, is the current humidity of the room, is the historical occupancy pattern characteristics; For short-term outings, the duration of the short-term absence is calculated by combining the access control and device status. The calculation formula is: , in, The duration of a user's brief absence. Using the random forest model, The time of the most recent access control operation. is the probability of leaving for a short time, Setting time thresholds : , , Among them, is the temperature control threshold for short-term going out, set to 10 minutes - 20 minutes.

[0017] The beneficial effects of the present invention are as follows: In the present invention, by combining the long short-term memory network (LSTM) with Bayesian optimization, collecting building environment, room heating and cooling demands, and historical energy consumption data, a short-term and long-term cooling load prediction model is constructed to dynamically adjust the cold and hot water supply strategy; the cold and hot water supply adjustment part is based on the cooling optimization parameters, combined with variable frequency water pump control, optimization of the water supply flow rate on each floor, and balanced adjustment of the cold and hot water pipe network, so that the cold and hot water delivery precisely matches the demands of different rooms, alleviating the situation of insufficient cooling on high floors and excessive cooling on low floors.

[0018] In the present invention, an error feedback correction mechanism is introduced to monitor the deviation between the predicted value and the actual load in real time. If the error exceeds the threshold, the LSTM model is automatically adjusted and the cooling parameters are updated to improve the prediction accuracy, reduce the cooling and heating load regulation error, and reduce the impact of temperature fluctuations on the user experience; for the temperature control requirements of different rooms, based on the real-time temperature, humidity, air velocity, and user preferences, the room temperature control strategy is calculated to appropriately increase the cooling in the sunny rooms at noon and reduce the cooling in the shady rooms.

[0019] In the present invention, the LSTM is combined with logistic regression analysis to analyze the reservation system, access control data, and equipment usage records to judge the actual usage status of the room and intelligently adjust the air conditioner operation mode; for empty rooms, the air conditioner maintains low-power operation to reduce energy waste; 10 - 20 minutes before the guests check in, the cooling is automatically started; for a short-term departure from the room, the departure duration is predicted through access control and equipment usage data to avoid unnecessary energy consumption.

[0020] In summary, the present invention can precisely match the heating and cooling demands of each room, improve the system response speed, reduce energy waste, make the cold and hot water supply more reasonable, improve the user comfort, and at the same time reduce the overall energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings 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.

[0022] Figure 1 It is a schematic flow chart of the building energy management method based on big data of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific implementation manners of the present invention in conjunction with the accompanying drawings of the specification.

[0024] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0025] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0026] Example 1, referring to Figure 1 , this example provides a building energy management method based on big data, including the following steps: Step S1, collect building environment data and room heating and cooling demand data, and monitor operating parameters; The building environment data includes weather change information, room orientation information, and external temperature and humidity data; The room heating and cooling demand data includes room occupancy rate, historical temperature preferences of occupants, and energy consumption of internal equipment in the room; The operating parameters include the current cooling capacity of the central air-conditioning system, the pressure of the chilled water supply and return, and the room temperature set value; Step S2, based on the data collected in step S1, calculate the predicted value of the heating and cooling load, and generate cooling optimization parameters; The predicted value of the heating and cooling load predicts the future cooling load demand through the long short-term memory network LSTM and the Bayesian optimization algorithm, and adjusts the cooling strategy of the central air-conditioning system; The cooling optimization parameters include the predicted cooling load demand of each room, the adjustment amount of the chilled water supply, and the target water supply pressure of the variable-frequency water pump; In step S2, the method of predicting the future cooling load demand through the long short-term memory network LSTM and the Bayesian optimization algorithm includes: Collect data of the past time periods as training data, including environmental variables, historical loads, occupancy situations, etc. The time period is set to 10 - 20 minutes, and the total time span is set to 24 hours or 48 hours. The data matrix is expressed as: , where, is the training data matrix, is the feature vector at the th time step, is the time step index, is the total number of time steps of the training data; Perform model training of LSTM. LSTM uses time series modeling, and the input is historical data , and the output is the future cooling load demand , , wherein, is the LSTM hidden state, is the hidden state at the previous time step, is the hidden state weight matrix, is the input weight matrix, is the input data, is the hidden state bias, is the activation function, and the hyperbolic tangent function is used here, is the predicted cooling load, is the output weight matrix, is the output bias; Adopt Gaussian process regression GPR for Bayesian optimization, with the goal of minimizing the prediction error. The optimization process is expressed as: , wherein, is the loss function, is the cooling load predicted by the model, is the actual cooling load, Bayesian optimization calculates the optimal value of the hyperparameter , and the calculation formula is: , wherein, is the set of hyperparameters to be optimized, represents the expectation of the loss function, is the optimized hyperparameter; In step S2, in the step of adjusting the cooling strategy of the central air-conditioning system, calculate the cooling optimization parameters, including: The room cooling load demand, and the calculation formula is: , wherein, is the room cooling load demand, is the empirical regression function, and multivariate linear regression is used for modeling, is the number of occupants, is the indoor temperature, is the outdoor temperature, is the relative air humidity, The adjustment amount of cold and hot water supply, and the calculation formula is: , where, is the adjustment amount of cold and hot water supply, is an empirical function and is modeled using logistic regression, is the current water supply pressure, is the water supply temperature, The target water supply pressure of the water pump, and the calculation formula is: , where, is the target water supply pressure, is the pressure prediction function and is calculated using support vector regression SVR, is the floor height factor; Introduce a dynamic error correction mechanism: , If , then automatically adjust the LSTM model and re-optimize the cooling parameters, where, is the relative error, is the predicted cooling load, is the actual cooling load, is the error threshold, which is set to 2% - 5%; Specifically, in step S2, the long short-term memory network LSTM is used for cooling load prediction, and the hyperparameters are adjusted in combination with Bayesian optimization to improve the prediction accuracy; specifically: Collect historical data and input it into the LSTM for time series modeling to calculate the future cooling load demand; use Bayesian optimization to adjust the LSTM structure to minimize the prediction error. Based on the prediction results, calculate the cooling load demand of each room and adjust the cooling strategy of the central air conditioner, including room cooling load calculation, adjustment amount calculation of cold and hot water supply, and calculation of the target water supply pressure of the water pump, to achieve precise cooling; at the same time, introduce a dynamic feedback correction mechanism to calculate the deviation error between the actual load and the predicted value. If the error exceeds the set threshold, then re-adjust the LSTM model and update the cooling optimization parameters; Here, a dynamic cooling strategy adjustment mechanism is provided by combining short-term and long-term cooling load predictions. At the same time, Bayesian optimization is used to reduce the model error, improve the cooling efficiency of the central air conditioner system, and optimize energy management; Step S3, according to the cooling optimization parameters generated in step S2, adjust the cold and hot water supply and calculate the flow regulation data; In step S2, the adjustment of cold and hot water supply includes dynamic regulation of the water supply pressure of the variable-frequency water pump, optimization of the water supply flow rate on each floor, and dynamic equilibrium control of the cold and hot water pipe network; The flow control data includes the set value of the water supply flow rate for each room and the set value of the water supply pressure of the water pump on different floors; In step S3, the methods for dynamic regulation of the water supply pressure of the variable-frequency water pump, optimization of the water supply flow rate on each floor, and dynamic equilibrium control of the cold and hot water pipe network include: Set the target water supply pressure , calculate the water pump power to match the cooling load demand on each floor, and the formula is: , where, is the water pump power, is the current water supply flow rate, is the target water supply pressure, is the characteristic coefficient of the water pump system, calibrated based on the pump efficiency curve; Optimize the water supply flow rate on each floor according to the occupancy rate, cooling and heating load demand, and historical data on each floor , and the optimization formula is: , where, is the water supply flow rate on the floor, is predicted using a multiple linear regression model, is the number of currently occupied rooms on the floor, is the floor height, is the outdoor temperature; Perform dynamic equilibrium on the cold and hot water pipe network, and dynamically adjust the cold and hot water supply based on the temperature difference between supply and return water and the flow feedback data. The adjustment formula is: , where, is the water supply pressure adjustment amount, and here is adjusted using PID control, is the return water temperature, is the return water flow rate; Finally, determine the time step for dynamic adjustment of the cold and hot water supply, and the determination formula is: , where, is the water supply adjustment time interval, uses an adaptive time step algorithm, is the number of currently occupied rooms on the current floor, is the outdoor temperature, is the pressure adjustment amount; Specifically, the target water supply pressure is calculated based on the cooling load requirements of each floor to adjust the pump power, matching the pressure with the load. The pump power is optimized according to the flow-pressure relationship to ensure the conveying efficiency. Secondly, the optimal water supply flow is calculated based on the floor occupancy rate and the cooling and heating load requirements to make the water supply match the actual demand; Here, multiple regression analysis is used for flow distribution to ensure the balance of cold and hot water supply to each floor. At the same time, the return water temperature and flow feedback control are introduced. The pressure adjustment amount is calculated through PID control, and the cold and hot water pipe network is dynamically adjusted to maintain the stability of the temperature and flow distribution on each floor; Step S4: Based on the flow regulation data calculated in step S3, execute room temperature control and calculate the room temperature control strategy; Room temperature control includes real-time monitoring of the room temperature and humidity status, air velocity analysis, and identification of the temperature preferences of the occupants; The room temperature control strategy includes increasing the cooling capacity during the high-temperature period at noon for sunny rooms, reducing the cooling capacity for shady rooms, and adjusting the target temperature in combination with the personalized settings of the occupants; Step S5: Based on the room temperature control strategy calculated in step S4, link the occupancy status data and calculate the final air-conditioning operation mode; The occupancy status data includes the room occupancy situation, the status of the reservation system, access control data, and the usage records of room equipment; The air-conditioning operation mode includes maintaining low-power operation at 26°C - 28°C for unoccupied rooms, starting to cool down 10 to 30 minutes before the guests check in, reducing the cold supply when leaving the room for a short time, and restoring the set temperature when returning; In step S5, the steps of linking the occupancy status data and calculating the final air-conditioning operation mode are as follows: Collect the reservation system, access control data, and usage records of room equipment, conduct occupancy situation analysis, and construct a data matrix, expressed as: , where, is the occupancy status data set, is the occupancy status data at the th time step, is the time span of the occupancy data, is the time step index; The occupancy status data includes: Reservation information: reservation check-in time and early check-out situation, Access control data: time series of entering and leaving the room, Equipment usage: usage situations of indoor facilities such as air conditioners and lights; The long short-term memory network LSTM and logistic regression LR are used to identify the user pattern, and the identification formula is: , wherein, is the probability of the user's short-term departure, is the sigmoid activation function, is the pattern recognition weight matrix, is the occupancy dataset, is the bias term; In step S5, the step of linking the occupancy status data and calculating the final air-conditioning operation mode further includes, starting the air conditioner minutes before the guest is expected to check in, and calculating the lead time: , wherein, is the air-conditioner early start time, adopting a linear regression model, is the outdoor temperature, is the current humidity of the room, is the historical occupancy pattern feature; For the short-term going-out scenario, the short-term departure duration is calculated by combining the access control and device status, and the calculation formula is: , wherein, is the short-term departure duration of the user, where adopting a random forest model, is the time of the most recent access control operation, is the short-term departure probability, setting a time threshold : , , wherein, is the short-term going-out temperature control threshold, set to 10 minutes - 20 minutes; Specifically, in step S5, the optimal air-conditioning operation mode is calculated by combining the occupancy status data, user pattern prediction, and intelligent cooling strategy, so as to achieve personalized energy-saving control; Based on the occupancy data, LSTM and logistic regression are used to predict the user pattern, and judge the short-term or long-term departure situation of the user; in the air-conditioning startup strategy, calculate the cooling timing before the user is expected to check in; at the same time, for the short-term departure situation, combine the access control data and device status to calculate the departure duration of the user, and judge whether it is necessary to reduce the cooling capacity to reduce energy consumption; Thereby, the user behavior is intelligently identified and the air-conditioning operation is dynamically adjusted, minimizing energy waste on the premise of ensuring comfort, introducing a machine learning prediction model, and dynamically adapting to the personalized needs of different users; Step S6: According to the air-conditioning operation mode calculated in step S5, perform the regulation and control of the central air-conditioning system to make the cold and hot water delivery match the needs of each room and conduct cold and heat load distribution.

[0027] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than 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 management method based on big data, characterized in that: including Step S1: Collect building environment data and room cooling and heating demand data, and monitor operating parameters; Step S2: Based on the data collected in Step S1, calculate the predicted value of the cooling and heating load, and generate cooling optimization parameters; Step S3: According to the cooling optimization parameters generated in Step S2, adjust the supply of chilled and hot water, and calculate the flow regulation data; In Step S2, the adjustment of the supply of chilled and hot water includes dynamic adjustment of the water supply pressure of the variable-frequency water pump, optimization of the water supply flow rate on each floor, and dynamic balance control of the chilled and hot water pipe network; The flow regulation data includes the set value of the water supply flow rate for each room and the set value of the water supply pressure of the water pump on different floors; Step S4: Based on the flow regulation data calculated in Step S3, perform room temperature control, and calculate the room temperature control strategy; Step S5: According to the room temperature control strategy calculated in Step S4, link the occupancy status data, and calculate the final air-conditioning operation mode; Step S6: According to the air-conditioning operation mode calculated in Step S5, execute the central air-conditioning system regulation, so that the delivery of chilled and hot water matches the needs of each room, and perform cooling and heating load distribution.

2. A building energy management method based on big data according to claim 1, characterized in that: The building environment data includes weather change information, room orientation information, and external temperature and humidity data; The room cooling and heating demand data includes room occupancy rate, historical temperature preferences of occupants, and energy consumption of internal equipment in the room; The operating parameters include the current cooling capacity of the central air-conditioning system, the chilled water supply and return water pressure, and the room temperature set value.

3. A building energy management method based on big data according to claim 2, characterized in that: The predicted value of the cooling and heating load predicts the future cooling load demand through the long short-term memory network LSTM and the Bayesian optimization algorithm, and adjusts the cooling strategy of the central air-conditioning system; The cooling optimization parameters include the predicted cooling load demand of each room, the adjustment amount of the chilled and hot water supply, and the target water supply pressure of the variable-frequency water pump.

4. The method for building energy management based on big data according to claim 3, wherein: In Step S2, the method for predicting the future cooling load demand through the long short-term memory network LSTM and the Bayesian optimization algorithm includes: Collect the data of the past for the following number of time periods as training data, where the time period is set to 10 - 20 minutes, and the total time span is set to 24 hours or 48 hours. The data matrix is represented as: ， Among them, is the training data matrix, is the feature vector at the -th time step, is the time step index, is the total number of time steps of the training data; Perform the model training of LSTM. LSTM uses time series modeling, and the input is historical data , and the output is the future cooling load demand . The calculation process includes: ， ， Among them, is the LSTM hidden state, is the hidden state of the previous time step, is the hidden state weight matrix, is the input weight matrix, is the input data, is the hidden state bias, is the activation function, and the hyperbolic tangent function is adopted here, is the predicted cooling load, is the output weight matrix, is the output bias; Using Gaussian process regression GPR for Bayesian optimization, with the goal of minimizing the prediction error, and the optimization process is expressed as: ， Among them, is the loss function, is the model's predicted cooling load, is the actual cooling load, Bayesian optimization calculates the optimal value of the hyperparameters using the following formula: ， Among them, is the set of hyperparameters to be optimized, represents the expectation of the loss function, is the optimized hyperparameter.

5. The building energy management method based on big data according to claim 4, characterized in that: In Step S2, in the step of adjusting the cooling strategy of the central air-conditioning system, calculating the cooling optimization parameters includes: Room cooling load demand, the calculation formula is: ， Among them, is the room cooling load demand, is an empirical regression function, and multiple linear regression is used for modeling, is the number of occupants, is the indoor temperature, is the outdoor temperature, is the relative air humidity, Adjustment amount of chilled and hot water supply, the calculation formula is: ， Among them, is the adjustment amount of cold and hot water supply, is an empirical function, and is modeled using logistic regression, is the current water supply pressure, is the water supply temperature, Target water supply pressure of the water pump, the calculation formula is: ， Among them, is the target water supply pressure, is the pressure prediction function, calculated using Support Vector Regression (SVR), is the floor height factor; Introduce a dynamic error correction mechanism: ， If , the LSTM model is automatically adjusted and the cooling parameters are re-optimized, where is the relative error, is the predicted cooling load, is the actual cooling load, is the error threshold, set to 2% - 5%.

6. The building energy management method based on big data according to claim 5, characterized in that: In Step S3, the methods for dynamic adjustment of the water supply pressure of the variable-frequency water pump, optimization of the water supply flow rate on each floor, and dynamic balance control of the chilled and hot water pipe network include: Set the target water supply pressure , calculate the pump power to match the cooling load requirements of each floor, and the formula is: ， Among them, is the pump power, is the current water supply flow rate, is the target water supply pressure, is the pump system characteristic coefficient, calibrated based on the pump efficiency curve; Optimize the water supply flow rate for each floor according to the occupancy rate, heating and cooling load requirements, and historical data of each floor , and the optimization formula is: ， Among them, is the water supply flow rate for the floor, predicted using a multiple linear regression model, is the current number of occupied rooms on the floor, is the floor height, is the outdoor temperature; Perform dynamic balance on the chilled and hot water pipe network, and dynamically adjust the chilled and hot water supply based on the supply and return water temperature difference and flow feedback data. The adjustment formula is: ， Among them, is the water supply pressure adjustment amount, where PID control is adopted for adjustment, is the return water temperature, is the return water flow rate; Finally, determine the time step of the dynamic adjustment of the chilled and hot water supply, and the determination formula is: ， Among them, is the water supply adjustment time interval, adopts an adaptive time step algorithm, is the number of occupied rooms on the current floor, is the outdoor temperature, is the pressure adjustment amount.

7. A building energy management method based on big data according to claim 6, characterized in that: The room temperature control includes real-time monitoring of the room temperature and humidity status, air velocity analysis, and identification of the temperature preferences of occupants; The room temperature control strategy includes increasing the cooling capacity during the high-temperature period at noon for sunny rooms, reducing the cooling capacity for shady rooms, and adjusting the target temperature in combination with the personalized settings of the guests.

8. A building energy management method based on big data according to claim 7, characterized in that: The occupancy status data includes the room occupancy situation, the status of the reservation system, access control data, and the usage records of room equipment; The air-conditioning operation mode includes maintaining a low-power operation at 26°C - 28°C for unoccupied rooms, starting to cool down 10 - 20 minutes before the guests check in, and reducing the cold supply when leaving the room for a short time and restoring the set temperature when returning.

9. The building energy management method based on big data according to claim 8, characterized in that: In step S5, the step of associating the occupancy status data and calculating the final air-conditioning operation mode is as follows: Collect the reservation system, access control data, and room equipment usage records, conduct an occupancy situation analysis, and construct a data matrix, expressed as: ， Among them, is the occupancy status data set, is the occupancy status data at the th time step, is the time span of the occupancy data, is the time step index; The occupancy status data includes: Reservation information: the reserved check-in time and early check-out situation, Access control data: the time series of entering and leaving the room, Equipment usage: the usage situations of air conditioners and lights; Use the long short-term memory network LSTM and logistic regression LR to identify user patterns. The identification formula is: ， Among them, is the probability of the user's short-term departure, is the sigmoid activation function, is the pattern recognition weight matrix, is the occupancy dataset, is the bias term.

10. The building energy management method based on big data according to claim 9, characterized in that: In step S5, the step of associating the occupancy status data and calculating the final air-conditioning operation mode further includes: Turn on the air conditioner minutes before the guest is expected to check in and calculate the advance time: ， Among them, is the early start time of the air conditioner, adopts a linear regression model, is the outdoor temperature, is the current humidity of the room, is the historical occupancy pattern feature; For the short-term going-out scenario, calculate the short-term leaving duration in combination with the access control and equipment status. The calculation formula is: ， Among them, is the duration of the user's short-term absence. Here, a random forest model is adopted, is the time of the most recent access control operation, is the probability of short-term absence. Set time threshold : ， ， Among them, is the temperature control threshold for a short-term absence, set to 10 minutes - 20 minutes.

Images