An adaptive energy consumption optimization control method and system of an intelligent building system

Abstract

This invention discloses an adaptive energy consumption optimization control method and system for intelligent building systems. The method includes: real-time collection of personnel density data, outdoor meteorological parameters, and real-time electricity price information for various areas within the building to generate a multi-source dynamic environmental feature set; outputting a regional load demand distribution map based on the multi-source dynamic environmental feature set; generating an initial collaborative parameter set for lighting, air conditioning, and security subsystems based on the regional load demand distribution map and real-time electricity price information; performing dynamic compensation for building thermal inertia on the initial collaborative parameter set to generate a transition parameter set corrected for thermal inertia; performing coupling degree analysis and conflict resolution between subsystems on the transition parameter set to output a conflict-free subsystem collaborative operation parameter set, which is then distributed to the lighting controller, air conditioning controller, and security controller for execution. Using this invention, adaptive collaborative optimization between building subsystems can be achieved, effectively reducing total building energy consumption and improving operational economy.

Description

Technical Field

[0001] This invention belongs to the field of intelligent building management technology, and in particular to an adaptive energy consumption optimization control method and system for intelligent building systems. Background Technology

[0002] Current intelligent building energy management often employs fixed rule-based or single-system independent control methods, making it difficult to achieve dynamic collaborative optimization across subsystems such as lighting, air conditioning, and security. Traditional methods often neglect the integration and utilization of multi-source dynamic information such as real-time occupant distribution, weather changes, and electricity price periods, resulting in insufficient accuracy in cooling and heating load forecasting and illuminance demand estimation. Furthermore, existing systems rarely consider the thermal inertia characteristics of the building envelope, and air conditioning control lacks dynamic compensation for heat storage delays and thermal comfort fluctuations, leading to unreasonable pre-cooling / pre-heating start / stop timing and supply air temperature settings, easily causing energy waste or decreased comfort. Simultaneously, there are often coupling conflicts between subsystem operating parameters, such as inconsistencies between lighting requirements and security linkage requirements, lacking effective conflict resolution and priority arbitration mechanisms. Summary of the Invention

[0003] The purpose of this invention is to provide an adaptive energy consumption optimization control method and system for intelligent building systems, so as to overcome the shortcomings of the prior art, realize adaptive collaborative optimization between building subsystems, effectively reduce the total energy consumption of the building and improve the economic efficiency of operation.

[0004] One embodiment of this application provides an adaptive energy consumption optimization control method for an intelligent building system, the method comprising: Real-time data collection of personnel density, outdoor meteorological parameters, and real-time electricity price information in various areas of the building is used to generate a multi-source dynamic environmental feature set. Based on the multi-source dynamic environmental feature set, the hourly cooling and heating loads and illuminance requirements of each functional area within a preset time period are calculated using a machine learning prediction model, and a regional load demand distribution map is output. Based on the regional load demand distribution map and the real-time electricity price time information, a multi-objective optimization function with the goal of minimizing total energy consumption is constructed to generate the initial collaborative parameter set for lighting, air conditioning and security subsystems. The initial collaborative parameter set is dynamically compensated for by building thermal inertia. Based on the heat storage delay characteristics of the building envelope and the allowable fluctuation range of indoor thermal comfort, the precooling and preheating start and stop times and supply air temperature setting curves of the air conditioning system are adjusted to generate a transition parameter set corrected by thermal inertia. The set of transition parameters is subjected to inter-system coupling analysis and conflict resolution. The conflicting operating parameters are adjusted through a fuzzy priority arbitration mechanism, and a set of conflict-free subsystem collaborative operating parameters is output and sent to the lighting controller, air conditioning controller and security controller for execution, so as to realize the adaptive energy consumption optimization control of the intelligent building system.

[0005] Optionally, the real-time collection of personnel density data, outdoor meteorological parameters, and real-time electricity price information in various areas of the building to generate a multi-source dynamic environmental feature set includes: Real-time statistics on the number of people entering and exiting each functional area and calculation of instantaneous personnel density are generated to produce time-series data on regional personnel density. Collect outdoor temperature, humidity, solar radiation intensity and wind speed data, and simultaneously obtain real-time electricity price information to generate raw outdoor weather and electricity price datasets; The time-series data of regional population density and the raw data of outdoor weather and electricity prices are time-stamped and aligned. The missing sampling points are filled in by linear interpolation to generate a time-synchronized multi-source data matrix. The time-synchronized multi-source data matrix is ​​normalized and feature-encoded, and the electricity price period is converted into a tiered numerical label, ultimately generating a multi-source dynamic environment feature set.

[0006] Optionally, the step of calculating the hourly cooling and heating loads and illuminance requirements of each functional area within a preset future time period based on the multi-source dynamic environmental feature set and using a machine learning prediction model, and outputting a regional load demand distribution map, includes: The multi-source dynamic environment feature set is divided into an input sliding window sequence according to the time order, and the feature data from one hour before the current time to the current time are extracted to generate the prediction input tensor. The prediction input tensor is input into a pre-trained long short-term memory network model. This model has learned the mapping relationship between historical load data and environmental characteristics, and outputs the predicted cooling load and heating load of each functional area in the future preset time period, generating a cooling and heating load prediction matrix. A multilayer perceptron model is used to process the nonlinear relationship between personnel density and illuminance demand. Combined with outdoor light intensity data, the illuminance demand values ​​of each functional area in the future preset time period are predicted, and an illuminance demand prediction vector is generated. The heating and cooling load prediction matrix and the illuminance demand prediction vector are spatially overlaid and visualized to generate a color heat map that identifies the hourly load values ​​of each functional area, and finally outputs a regional load demand distribution map.

[0007] Optionally, the step of constructing a multi-objective optimization function with the goal of minimizing total energy consumption based on the regional load demand distribution map and the real-time electricity price time information, and generating an initial set of coordinated parameters for the lighting, air conditioning, and security subsystems, includes: Analyze the hourly heating and cooling loads and illuminance demand in the load demand distribution map, combine the peak and valley periods of electricity price to determine the energy cost sensitivity of each period, and generate an optimization decision boundary with priority weights. Based on the optimized decision boundary, lighting energy consumption models, air conditioning energy consumption models, and security energy consumption models are established respectively, and subsystem energy consumption mapping relationships are generated. By utilizing the energy consumption mapping relationship of subsystems, a multi-objective optimization function is constructed with the goal of minimizing total energy consumption and constraints of thermal comfort, illuminance satisfaction, and security response coverage. Electricity price weighting factors are introduced to generate the optimization objective function. Solve the objective function to obtain recommended operating parameters for each subsystem, including at least the air conditioning set temperature and start / stop time, lighting dimming curve, security inspection frequency and hibernation strategy, and generate the initial collaborative parameter set for the lighting, air conditioning and security subsystems.

[0008] Optionally, the step of performing dynamic compensation for building thermal inertia on the initial collaborative parameter set, adjusting the precooling and preheating start / stop times and supply air temperature set curves of the air conditioning system according to the heat storage delay characteristics of the building envelope and the allowable fluctuation range of indoor thermal comfort, and generating a transition parameter set corrected for thermal inertia includes: Extract the building envelope material parameters of each functional area from the building information model, including wall thickness, density, specific heat capacity and thermal conductivity, and generate a table of thermal property parameters of the building envelope. Based on the thermal property parameter table of the building envelope, the one-dimensional unsteady heat conduction equation is solved by the finite difference method, the response delay time of indoor temperature to air conditioning start-up and shutdown is calculated, and the heat storage delay characteristic curves of each zone are generated. Based on the heat storage delay characteristic curve and the allowable fluctuation range of indoor thermal comfort, the start and stop times of the air conditioner in the initial coordination parameter set are adjusted in advance or in advance, and the supply air temperature setting curve is recalculated to generate a pre-cooling and preheating adjustment scheme. The precooling and preheating adjustment scheme is merged into the lighting and security parameters, the start and stop times and supply air temperature set values ​​of the air conditioning section are updated, and a set of transition parameters corrected by thermal inertia is generated.

[0009] Optionally, the step of performing inter-subsystem coupling analysis and conflict resolution on the transition parameter set, adjusting contradictory operating parameters through a fuzzy priority arbitration mechanism, outputting a conflict-free subsystem collaborative operating parameter set, and distributing it to the lighting controller, air conditioning controller, and security controller for execution, thereby realizing adaptive energy consumption optimization control of the intelligent building system, includes: The operation parameters of the three subsystems of centralized lighting, air conditioning and security are analyzed in the transition parameters, the coupling relationship between the parameters is identified, and the coupling influence factor matrix is ​​generated. Conflicting parameter combinations are detected based on the coupling influence factor matrix, conflicting parameter pairs are marked and conflict degree coefficients are calculated to generate a list of conflicting parameters. Based on the fuzzy priority arbitration mechanism, priority weights are assigned to each subsystem according to time periods, and conflicting parameters are adjusted by compromise according to weights to generate a set of parameters for the coordinated operation of non-conflicting subsystems. The set of parameters for the coordinated operation of conflict-free subsystems is distributed to the lighting controller, air conditioning controller and security controller for execution, and the execution status is returned to realize the adaptive energy consumption optimization control of the intelligent building system.

[0010] Another embodiment of this application provides an adaptive energy consumption optimization control system for an intelligent building system, the system comprising: The data acquisition module is used to collect real-time data on personnel density, outdoor meteorological parameters, and real-time electricity price information in various areas of the building, and generate a multi-source dynamic environmental feature set. The prediction module is used to calculate the hourly cooling and heating loads and illuminance requirements of each functional area within a preset time period based on the multi-source dynamic environmental feature set and through a machine learning prediction model, and output a regional load demand distribution map. The construction module is used to construct a multi-objective optimization function with the goal of minimizing total energy consumption based on the regional load demand distribution map and the real-time electricity price time information, and to generate the initial collaborative parameter set of lighting, air conditioning and security subsystems; The adjustment module is used to perform dynamic compensation for building thermal inertia on the initial collaborative parameter set. Based on the heat storage delay characteristics of the building envelope and the allowable fluctuation range of indoor thermal comfort, the module adjusts the precooling and preheating start and stop times and supply air temperature setting curves of the air conditioning system to generate a transition parameter set corrected for thermal inertia. The control module is used to perform inter-subsystem coupling analysis and conflict resolution on the transition parameter set, adjust the contradictory operating parameters through a fuzzy priority arbitration mechanism, output a set of conflict-free subsystem collaborative operating parameters, and send them to the lighting controller, air conditioning controller and security controller for execution, so as to realize the adaptive energy consumption optimization control of the intelligent building system.

[0011] Another embodiment of this application provides a storage medium storing a computer program, wherein the computer program is configured to execute the method described in any of the preceding claims when running.

[0012] Another embodiment of this application provides an electronic device including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the method described in any of the preceding claims.

[0013] Compared with existing technologies, the adaptive energy consumption optimization control method for intelligent building systems provided by this invention can realize adaptive collaborative optimization among building subsystems, effectively reduce total building energy consumption and improve operational economy. Attached Figure Description

[0014] Figure 1 A hardware structure block diagram of a computer terminal for an adaptive energy consumption optimization control method for an intelligent building system provided in an embodiment of the present invention; Figure 2 A flowchart illustrating an adaptive energy consumption optimization control method for an intelligent building system provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an adaptive energy consumption optimization control system for an intelligent building system provided in an embodiment of the present invention. Detailed Implementation

[0015] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0016] This invention first provides an adaptive energy consumption optimization control method for intelligent building systems. This method can be applied to electronic devices, such as computer terminals, specifically ordinary computers.

[0017] The following detailed explanation uses a computer terminal as an example. Figure 1 This is a hardware structure block diagram of a computer terminal for an adaptive energy consumption optimization control method for an intelligent building system provided in an embodiment of the present invention. (See diagram below.) Figure 1 As shown, the computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include non-volatile storage media and internal memory.

[0018] See Figure 2 The present invention provides an adaptive energy consumption optimization control method for an intelligent building system, which may include the following steps: S201 collects real-time data on personnel density, outdoor meteorological parameters, and real-time electricity price information in various areas of the building, and generates a multi-source dynamic environmental feature set. Specifically, it can count the number of people entering and exiting each functional area in real time and calculate the instantaneous population density to generate time-series data on population density in the area; The core of this step is to accurately acquire the flow of people in various functional areas of the building through multi-source sensing devices, obtain the instantaneous population density through quantitative calculation, and store it in a time series to form continuous time-series data, providing support for subsequent load forecasting based on population activity characteristics. The specific implementation method is as follows: The building's interior is divided into several independent functional areas, including offices, meeting rooms, a lobby, corridors, and restrooms. Each functional area is equipped with corresponding personnel sensing devices, primarily including two-way infrared passenger flow sensors, millimeter-wave human presence radar, and intelligent video people counting cameras. These three types of devices work together to avoid errors from individual devices and ensure the accuracy of headcount statistics. The two-way infrared passenger flow sensors are installed above the door frames at the entrances and exits of the functional areas, accurately distinguishing between people entering and leaving with a counting accuracy of over 98%, updating the count every second. The millimeter-wave human presence radar is deployed on the ceiling inside the functional areas to monitor the number of people staying in the area in real time, compensating for the sensor's limitation of only counting entrances and exits, and covering open areas without clearly defined entrances and exits. The intelligent video people counting cameras use AI human detection algorithms to visually count people in the area, suitable for large open functional areas, further improving the accuracy of the statistics.

[0019] During real-time statistics, the system acquires data from each device at a fixed sampling period of 1 minute. This period accurately captures dynamic changes in personnel density without causing excessive data volume due to excessive sampling frequency, thus meeting the efficiency requirements of subsequent data processing. For each functional area, the system accumulates the number of people entering and leaving from the bidirectional infrared sensors in real time. An internal algorithm calculates the total number of people in the area, using the formula: current total number of people equals accumulated number of people entering minus accumulated number of people leaving. Combined with the fixed area value of the functional area, the instantaneous personnel density is calculated. The unit of instantaneous personnel density is people per square meter, a core indicator reflecting the density of people in the area. A higher value indicates a greater demand for heating, cooling, and illumination in that area.

[0020] In the example, the office area is 100 square meters. At a certain moment, 52 people entered and 18 people left, with a current total of 34 people, resulting in an instantaneous personnel density of 0.34 people per square meter. The meeting room is 50 square meters, with a current total of 12 people, resulting in an instantaneous personnel density of 0.24 people per square meter. The system stores the instantaneous personnel density value of each functional area along with the corresponding sampling timestamp. The timestamp format is accurate to the minute, and the data is continuously recorded over time to form a time-series data sequence. The time-series data of each functional area is stored independently, including three core fields: area identifier, timestamp, and instantaneous personnel density. Finally, it generates regional personnel density time-series data covering all functional areas. This data completely records the changing pattern of personnel density in each area over time and is the core characteristic data reflecting the usage status inside the building.

[0021] Collect outdoor temperature, humidity, solar radiation intensity and wind speed data, and simultaneously obtain real-time electricity price information to generate raw outdoor weather and electricity price datasets; The core of this step is to collect external environmental parameters and energy cost parameters that affect building energy consumption, and to integrate outdoor meteorological data with real-time electricity price data to form a complete raw dataset of external conditions. This provides external input for subsequent load forecasting and energy consumption optimization. The specific implementation method is as follows: Outdoor meteorological parameters are collected through an integrated meteorological monitoring station deployed on the building facade. The station integrates four types of monitoring units: temperature sensors, humidity sensors, solar radiation intensity meters, and anemometers. All sensors employ industrial-grade precision to ensure stable and reliable data. The outdoor temperature sensor measures from -30°C to 60°C with an accuracy of ±0.1°C, collecting real-time atmospheric temperature data around the building; this is a core parameter affecting air conditioning heating and cooling loads. The outdoor humidity sensor measures from 0 to 100%RH with an accuracy of ±1%RH, reflecting the water vapor content in the air and relating to the building's dehumidification and humidification loads. The solar radiation intensity meter measures from 0 to 2000 watts per square meter with an accuracy of ±5 watts per square meter, accurately monitoring solar radiation intensity and directly affecting the building's heat gain and lighting loads. The anemometer measures from 0 to 60 meters per second with an accuracy of ±0.1 meters per second, reflecting outdoor airflow speed and affecting the heat exchange efficiency of the building envelope.

[0022] The sampling cycle of the meteorological monitoring station is consistent with the population density statistics cycle, both being 1 minute, ensuring the uniformity of the data time dimension. The four meteorological parameters collected are uploaded to the building energy consumption management system in real time without transmission delay. Real-time electricity price information is synchronously obtained through the power company's energy management platform. Currently, residential and commercial buildings generally implement a three-tiered pricing system based on peak, valley, and flat periods. The peak period is the time of highest electricity consumption with the highest price; the flat period is the time of stable electricity consumption with a moderate price; and the valley period is the time of lowest electricity consumption with the lowest price. The system obtains the current electricity price period and corresponding price value in real time, with the price value retained to two decimal places and the unit being yuan per kilowatt-hour.

[0023] In the example, the outdoor meteorological data collected at a certain moment is: temperature 28.5 degrees Celsius, humidity 55%RH, solar radiation intensity 750 watts per square meter, and wind speed 2.3 meters per second. The real-time electricity price acquired simultaneously is for peak hours, at 1.20 yuan per kilowatt-hour. The system integrates the four meteorological parameters—outdoor temperature, outdoor humidity, solar radiation intensity, and wind speed—with the real-time electricity price value and electricity price time type, and combines them uniformly by timestamp to form a single raw data record. Continuous data is generated over the sampling period, ultimately producing a raw outdoor meteorological and electricity price dataset containing timestamps, meteorological parameters, and electricity price information. This dataset comprehensively covers all external factors affecting building energy consumption, laying the foundation for multi-source data fusion.

[0024] The time-series data of regional population density and the raw data of outdoor weather and electricity prices are time-stamped and aligned. The missing sampling points are filled in by linear interpolation to generate a time-synchronized multi-source data matrix. The core of this step is to solve the problems of asynchronous time and missing sampling points in multi-source data. By aligning timestamps and filling in missing values, the population density, weather, and electricity price data are integrated into a regular matrix structure to ensure the consistency of subsequent data processing. The specific implementation method is as follows: The regional population density time-series data, outdoor weather data, and electricity price data come from different acquisition devices and data sources. Initially, there was an issue of incomplete timestamp matching. Timestamp alignment is a crucial step in achieving multi-source data fusion. The system uses a unified standard time as a benchmark, calibrating the timestamps of all data to the exact minute (00 seconds). It then compares the timestamps of population density data, weather data, and electricity price data one by one, matching and associating data of the same timestamp across different data types. This ensures that each time point corresponds to complete multi-source feature data, eliminating feature bias caused by time misalignment.

[0025] In actual data collection, due to factors such as equipment failure, network fluctuations, and signal interference, some timestamps may experience data loss. Linear interpolation is the optimal method for filling in these missing sampling points. This method is based on the continuous variation pattern of data, calculating the value of the missing point using the values ​​of two adjacent valid data points. The calculation logic is simple and the error is minimal, making it suitable for the stable variation characteristics of building environment data. The core principle of linear interpolation is that the value of the missing time point equals the value of the previous valid time point plus the difference between the values ​​of the next valid time point and the previous valid time point, multiplied by the ratio of the time interval between the two valid time points, ultimately yielding a smoothly transitioned missing value.

[0026] In the example, the population density data at 14:05 is missing. The population density at the previous valid time point of 14:04 was 0.32 people per square meter, and the population density at the next valid time point of 14:06 was 0.36 people per square meter. The two times are 2 minutes apart, and the missing time is in the middle. After interpolation, the population density at 14:05 is calculated to be 0.34 people per square meter. Similarly, missing points in meteorological parameters and electricity price data are filled using the same method to ensure that the data sequence is unbroken. After completing timestamp alignment and missing value filling, the system arranges all data in timestamp order to construct a multi-source data matrix. Each row of the matrix represents a timestamp, and each column represents a feature dimension, namely timestamp, office area population density, meeting room population density, lobby population density, outdoor temperature, outdoor humidity, solar radiation intensity, wind speed, and real-time electricity price. All data in the matrix are valid values ​​that are synchronized in time, without missing or misaligned values, forming a well-organized time-synchronized multi-source data matrix.

[0027] The time-synchronized multi-source data matrix is ​​normalized and feature-encoded, and the electricity price period is converted into a tiered numerical label, ultimately generating a multi-source dynamic environment feature set.

[0028] The core of this step is to eliminate dimensional differences through data standardization, transform categorical data into numerical data through feature encoding, and generate a standardized feature set that is suitable for machine learning models, providing qualified input for subsequent load forecasting. The specific implementation method is as follows: In a time-synchronized multi-source data matrix, the dimensions and numerical ranges of various feature data differ significantly. Human density ranges from 0 to 1 person per square meter, outdoor temperature from -30 to 60 degrees Celsius, solar radiation intensity from 0 to 2000 watts per square meter, and real-time electricity price from 0.4 to 1.2 yuan per kilowatt-hour. Directly using the raw data can lead to numerical imbalance during machine learning model training, affecting prediction accuracy. Therefore, normalization is necessary. This study employs a min-max normalization method, mapping all feature data to a unified range of 0 to 1. The normalization formula is: normalized value equals the original value minus the minimum value of the feature, divided by the difference between the maximum and minimum values ​​of the feature. After processing, all feature values ​​are on the same scale, eliminating the influence of dimensions and numerical ranges.

[0029] In the example, the original value of solar radiation intensity is 750 watts per square meter, with a minimum value of 0 and a maximum value of 2000, resulting in a normalized value of 0.375. The original value of outdoor temperature is 28.5 degrees Celsius, with a minimum value of -30 and a maximum value of +60, resulting in a normalized value of 0.65. All continuous features are standardized using this method. Feature encoding mainly targets the classification feature of electricity price periods, which are divided into peak, flat, and off-peak periods. These periods cannot be directly input into the numerical model, so they are converted into stepped numerical labels, with peak periods assigned a value of 3, flat periods 2, and off-peak periods 1. The stepped values ​​are positively correlated with the electricity price, thus preserving the classification attribute of the periods while transforming them into numerical features that the model can recognize.

[0030] After normalization and feature encoding, the system removes the timestamp field from the matrix, retaining only the standardized feature values, and integrates them to form a multi-source dynamic environmental feature set. This feature set is a purely numerical structure, containing three core dynamic features: personnel density, outdoor weather, and electricity price. All features have been standardized and encoded, and can be directly input into machine learning prediction models. The feature set is updated in real time with the collection cycle, dynamically reflecting the environmental changes inside and outside the building, and serves as the fundamental feature data for achieving adaptive energy consumption optimization in intelligent buildings.

[0031] S202, Based on the multi-source dynamic environmental feature set, calculate the hourly cooling and heating load and illuminance demand of each functional area within a future preset time period through a machine learning prediction model, and output a regional load demand distribution map. Specifically, the multi-source dynamic environment feature set can be divided into an input sliding window sequence according to the time order, and feature data from one hour before the current time to the current time can be extracted to generate a prediction input tensor. The core of this step is to perform time-dimensional sliding window segmentation and structured recombination of the multi-source dynamic environmental feature set, extract time-series feature data of fixed duration, and transform discrete environmental features into a standardized tensor structure that can be input into machine learning models, providing regular input data for subsequent hot and cold load prediction. The specific implementation method is as follows: The multi-source dynamic environmental feature set comprises multi-dimensional time-series data including population density, outdoor weather, and real-time electricity prices. Data is collected at a fixed sampling frequency with a 1-minute sampling interval to ensure the real-time and continuous nature of feature changes. Each data point in the feature set carries a precise timestamp, including year, month, day, hour, minute, and second, with timestamp accuracy down to the second level, guaranteeing the accuracy of time-series segmentation. The input sliding window sequence is a standardized data slicing format used for time-series forecasting. In this case, the sliding window spans from one hour prior to the current moment to the current moment, for a total duration of 60 minutes, corresponding to 60 consecutive sampled data points. The window ends at the current moment and slides and updates minute by minute over time, ensuring that the prediction input is always based on the latest environmental features.

[0032] The segmentation process iterates through the multi-source dynamic environmental feature set according to the chronological order of timestamps, selecting all data entries whose timestamps fall within the target time range. Abnormal entries with discontinuous timestamps or missing data are removed. For a small number of missing points, the mean of nearby timestamps is used to fill in the gaps, ensuring the integrity of the data within the window. The extracted feature data includes multi-dimensional feature items: instantaneous population density in each functional area, outdoor dry-bulb temperature, outdoor relative humidity, total solar radiation intensity, outdoor wind speed, wind direction, electricity price tier labels, weekday identifiers, and time period identifiers, totaling nine core environmental features covering all key factors affecting building heating and cooling loads and illuminance requirements.

[0033] Taking an office building with eight functional areas as an example, the extracted 60 time-series data points each contain personnel density values ​​for the eight functional areas and nine global environmental features, resulting in a total feature dimension of 17. The segmented two-dimensional time-series data is unfolded chronologically and reconstructed into a three-dimensional tensor structure. The first dimension of the tensor is the time step, with a value of 60, corresponding to a 60-minute time series length; the second dimension is the number of functional areas, with a value of 8, corresponding to the various functional zones within the building; and the third dimension is the feature dimension, with a value of 9, corresponding to various environmental feature parameters. This three-dimensional structured data serves as the prediction input tensor. The tensor values ​​have all undergone prior normalization, stabilizing their range between 0 and 1, eliminating prediction bias caused by differences in feature dimensions. This data can then be directly input into a long short-term memory network model for prediction calculations.

[0034] The prediction input tensor is input into a pre-trained long short-term memory network model. This model has learned the mapping relationship between historical load data and environmental characteristics, and outputs the predicted cooling load and heating load of each functional area in the future preset time period, generating a cooling and heating load prediction matrix. The core of this step is to leverage the time-series prediction advantages of the Long Short-Term Memory (LSTM) network model. Based on the mapping relationship trained on historical data, it achieves accurate prediction of future heating and cooling loads. The input tensor is transformed into quantified load prediction results, generating a standardized prediction matrix. The specific implementation method is as follows: The Long Short-Term Memory (LSTM) network model is a deep learning model adapted for time-series data prediction. It can effectively capture the long-term temporal dependencies between environmental features and building load, avoiding the gradient vanishing problem of traditional models. The model has been fully pre-trained before prediction. During the training phase, multi-source environmental feature data from the past three years of building history are used as input, and actual heating and cooling load monitoring data for the corresponding time periods are used as labels. An adaptive moment estimation optimization algorithm is used for training, with a learning rate of 0.001, a batch size of 32, and 100 training iterations. The final model's mean absolute percentage error on the validation set is controlled within 5%, meeting the accuracy requirements for building load prediction.

[0035] The core mapping relationships learned by the model include multiple key association rules. Increased population density leads to increased indoor heat dissipation, thereby increasing the demand for cooling load; increased outdoor temperature increases the building's cooling load and reduces the heating load; increased solar radiation intensity increases indoor heat gain, affecting the fluctuation of cooling and heating loads; and electricity price periods indirectly affect the optimal allocation of loads. The model can autonomously capture these nonlinear and temporal association features to ensure that the prediction results are consistent with the actual energy consumption patterns of buildings.

[0036] The future preset time period is set to the next 6 hours, which is a typical forward-looking time for building energy consumption optimization and control, balancing real-time control with forward-looking optimization. The model takes the prediction input tensor as input and outputs the predicted cooling and heating loads for each hour and each functional area within the next 6 hours, with the load unit being kilowatts and the prediction accuracy reaching 0.1 kilowatts. Taking an office building with 8 functional areas as an example, the model output prediction results include values ​​for the next 6 time steps, 8 functional areas, and 2 types of loads (cooling and heating). The results are arranged in a regularized manner according to functional areas and time steps to form a two-dimensional prediction matrix. The rows of the matrix correspond to the next 6 time points, and the columns correspond to the cooling and heating load values ​​for the 8 functional areas. Each matrix element represents the load prediction value for the corresponding time and corresponding functional area. This matrix is ​​the cooling and heating load prediction matrix. The matrix data is updated in real time, providing load basis data for subsequent illuminance prediction and visualization.

[0037] A multilayer perceptron model is used to process the nonlinear relationship between personnel density and illuminance demand. Combined with outdoor light intensity data, the illuminance demand values ​​of each functional area in the future preset time period are predicted, and an illuminance demand prediction vector is generated. The core of this step is to fit the nonlinear relationship between population density and illuminance demand using a multilayer perceptron model, integrate the natural lighting compensation effect of outdoor light intensity, predict illuminance demand synchronized with load periods, and generate a one-dimensional standardized prediction vector. The specific implementation method is as follows: The multilayer perceptron model is a feedforward neural network model that handles nonlinear mapping relationships. It consists of an input layer, three hidden layers, and an output layer. The hidden layers use rectified linear units as activation functions, which can effectively fit the complex nonlinear relationship between personnel density and illumination requirements. Personnel density is the core factor affecting illumination requirements. The higher the personnel density in functional areas such as offices and meeting rooms, the higher the required artificial lighting illumination to meet the needs of visual tasks. Outdoor light intensity provides natural lighting compensation; the higher the light intensity, the lower the required artificial lighting illumination. The model can simultaneously integrate these two key influencing factors to accurately predict illumination requirements.

[0038] The model's input data consists of the population density characteristics and outdoor light intensity characteristics of each functional area in the prediction input tensor. The output is the hourly illuminance demand value for each functional area over the next 6 hours, measured in lux, meeting the requirements of building lighting design standards, with a prediction accuracy of 10 lux. During the model training phase, historical population density and outdoor light intensity data are used as input, with actual illuminance monitoring values ​​as labels. After training, accurate end-to-end predictions are achieved, with the prediction period consistent with the heating and cooling load predictions, both set for the next 6 hours, ensuring data temporal synchronization.

[0039] Taking eight functional zones as an example, the model outputs illuminance demand values ​​for each functional zone at six future time points, totaling 48 predicted values. These values ​​are arranged sequentially by time and functional zone number, recombining them into a one-dimensional sequence of data. Each value corresponds to the illuminance demand of a specific functional zone at a specific time. This one-dimensional sequence of data is the illuminance demand prediction vector. The vector value range conforms to building illuminance code requirements, with illuminance demand in office areas maintained at 300 to 500 lux and in corridor areas at 100 to 200 lux. The model can automatically adjust the prediction results according to the functional zone type, ensuring that the illuminance demand matches the usage needs of different areas. The time step and functional zone number of the vector correspond perfectly with the heating and cooling load prediction matrix, laying a data foundation for subsequent spatial overlay.

[0040] The heating and cooling load prediction matrix and the illuminance demand prediction vector are spatially overlaid and visualized to generate a color heat map that identifies the hourly load values ​​of each functional area, and finally outputs a regional load demand distribution map.

[0041] The core of this step is to spatially match and visualize the predicted heating and cooling loads with illuminance demand data. A color heatmap visually displays the load distribution characteristics of each region, generating a regional load demand distribution map that can be directly used for optimization and control. The specific implementation method is as follows: Spatial overlay is the process of precisely matching time-series forecast data with the geospatial information of building functional areas. Based on the building's two-dimensional floor plan, the values ​​in the heating and cooling load forecast matrix and the illuminance demand forecast vector are bound one-to-one with the spatial coordinates of the corresponding functional areas, ensuring a precise correspondence between data and spatial location. During the overlay process, time steps are strictly aligned; heating and cooling load values ​​and illuminance demand values ​​at the same moment are overlaid onto the same functional area, forming a multi-indicator fused spatial data layer. The data overlay is seamless and complete, fully covering all functional areas of the building.

[0042] The visualization rendering employs a color-coded hierarchical mapping method, assigning differentiated color systems to different load indicators. Cooling load values ​​are mapped using a blue color scheme, with higher values ​​resulting in a deeper blue; heating load values ​​are mapped using a red color scheme, with higher values ​​resulting in a deeper red; and illuminance demand values ​​are mapped using a yellow color scheme, with higher values ​​resulting in a brighter yellow. Each color scheme is divided into 10 levels, corresponding to different numerical ranges, intuitively distinguishing differences in load intensity. The generated color heatmap uses the building's floor plan as a base map, with each functional area filled with a corresponding color, and hourly load values ​​are clearly labeled. The numerical accuracy remains consistent with the prediction results, and the view can be dynamically switched between six future time points to observe changes in load distribution over different periods.

[0043] The regional load demand distribution map is a visualization that integrates spatial information, temporal information, and multi-indicator load data. It includes five core elements: building floor plan, functional area division, hourly cooling and heating load values, hourly illuminance demand values, and color-coded legends. Presented in a lightweight vector format, it can be transmitted in real-time to the building energy consumption optimization control system. Taking office buildings as an example, the distribution map clearly shows the distribution characteristics of high cooling load and high illuminance demand in the office area during the morning, and the characteristics of low load and low illuminance demand in the corridor area. This provides an intuitive load basis for the subsequent construction of multi-objective optimization functions. The distribution map is updated every 5 minutes to ensure data real-time performance and effectiveness, supporting the precise execution of adaptive energy consumption optimization control.

[0044] S203, Based on the regional load demand distribution map and the real-time electricity price time information, construct a multi-objective optimization function with the goal of minimizing total energy consumption, and generate an initial set of collaborative parameters for lighting, air conditioning and security subsystems; Specifically, the hourly heating and cooling loads and illuminance demand in the load demand distribution map can be analyzed, and combined with the peak and valley periods of electricity prices, the energy cost sensitivity of each period can be determined, and an optimization decision boundary with priority weights can be generated. The core of this step is to extract quantified load data from a visualized regional load demand distribution map, calculate the energy cost sensitivity of each time period in conjunction with the time-of-use pricing system, define the threshold range for optimization control, and form a priority-based decision boundary. This provides accurate input for subsequent energy consumption modeling. The specific implementation method is as follows: The regional load demand distribution map is a color heat map that identifies the hourly load values ​​of each functional area. Different color gradients correspond to the magnitude of cooling and heating loads and illuminance requirements, with red representing high loads and blue representing low loads. Each functional area is marked with precise hourly values. The analysis process is completed by an image feature extraction and numerical recognition module. The module automatically traverses the coordinate areas of all functional areas in the heat map, such as office areas, meeting rooms, corridors, and computer rooms, and extracts the hourly cooling load, heating load, and illuminance requirements for the next 24 hours. The unit for cooling load is watts per square meter, the unit for heating load is also watts per square meter, and the unit for illuminance requirements is lux. The analysis accuracy reaches one data point per hour, and the numerical error is controlled within ±2 watts per square meter and ±5 lux. Finally, an hourly load demand data table is formed. In the example, the cooling load of the office area at 8:00 am is 85 watts per square meter and the illuminance requirement is 500 lux, and the cooling load of the meeting room at 2:00 pm is 110 watts per square meter and the illuminance requirement is 450 lux.

[0045] Real-time electricity pricing is divided into three periods according to the grid's time-of-use pricing rules: peak, normal, and off-peak. In the example, peak hours are 8:00-11:00 and 18:00-21:00, with a price of 1.2 yuan per kilowatt-hour; normal hours are 7:00-8:00, 11:00-18:00, and 21:00-22:00, with a price of 0.8 yuan per kilowatt-hour; and off-peak hours are from 22:00 to 7:00 the next day, with a price of 0.4 yuan per kilowatt-hour. Energy cost sensitivity is used to characterize the impact of electricity costs on total energy consumption during a given period. The sensitivity value is positively correlated with the electricity price and is calculated by dividing the current period's electricity price by the off-peak electricity price. In the example, the peak sensitivity is 3.0, the normal sensitivity is 2.0, and the off-peak sensitivity is 1.0. A higher value indicates a higher priority for energy conservation optimization during that period.

[0046] Priority weights are allocated based on energy cost sensitivity. Normalization is used to convert the sensitivity into weight values ​​between 0 and 1: 0.6 for peak hours, 0.3 for normal hours, and 0.1 for off-peak hours. These weights directly determine the priority of subsystem parameter adjustments during each time period. The optimization decision boundary is a control threshold range defined by combining load values ​​and electricity price weights. It includes upper and lower limits for cooling and heating load regulation, minimum guaranteed illuminance demand, and maximum allowable energy consumption for each time period. In the example, the upper limit for air conditioning cooling load regulation during peak hours is 80 watts per square meter, and the minimum guaranteed illuminance is 450 lux. The upper limit for cooling load regulation during off-peak hours can be relaxed to 100 watts per square meter. This boundary satisfies energy demand while defining constraints for energy consumption optimization, ultimately generating an optimization decision boundary that integrates load, electricity price, and weights.

[0047] Based on the optimized decision boundary, lighting energy consumption models, air conditioning energy consumption models, and security energy consumption models are established respectively, and subsystem energy consumption mapping relationships are generated. The core of this step is to establish energy consumption calculation models for the three subsystems of lighting, air conditioning, and security based on the constraints of the optimization decision boundary, clarify the correspondence between operating parameters and energy consumption values, and form a quantifiable subsystem energy consumption mapping relationship, providing a foundation for the construction of multi-objective optimization functions. The specific implementation method is as follows: The lighting energy consumption model is established based on the power characteristics of lighting equipment, dimming curves, personnel density, and illuminance requirements. The core logic is that lighting energy consumption equals the product of the functional area area, the lighting power per unit area, and the dimming rate. The dimming rate is dynamically adjusted according to illuminance requirements and outdoor light intensity. The minimum guaranteed illuminance value in the optimization decision boundary is the lower limit constraint of the model. In the example, the office area is 100 square meters, the lighting power per unit area is 10 watts per square meter, and the peak dimming rate is 0.5. At this time, the lighting energy consumption is 100 multiplied by 10 multiplied by 0.5, which equals 500 watts. The model also incorporates personnel density parameters. In unoccupied areas, the dimming rate drops to below 0.2, further reducing energy consumption. This model fully represents the mapping relationship between illuminance requirements, dimming curves, and lighting energy consumption.

[0048] The air conditioning energy consumption model is the most complex of the three subsystems. It is based on cooling and heating load, supply air temperature, start and stop times, room area, and building envelope heat transfer coefficient. The core logic is that air conditioning energy consumption equals the comprehensive calculation result of cooling and heating load, operating time, and energy efficiency ratio. The upper and lower limits of cooling and heating load regulation in the optimization decision boundary are the core constraints of the model. Energy consumption output is strictly controlled during peak hours with high electricity prices, while it can be appropriately relaxed during off-peak hours. In the example, the cooling load of the office area is 85 watts per square meter, the area is 100 square meters, the energy efficiency ratio is 3.5, and the energy consumption for one hour of operation is 85 multiplied by 100 divided by 3.5, which is approximately 2428 watts. The model also considers the impact of start and stop times on energy consumption. Starting and stopping earlier can reduce the operating time during peak hours. This model accurately represents the mapping relationship between cooling and heating load, operating parameters, and air conditioning energy consumption.

[0049] The security energy consumption model is based on the power consumption, inspection frequency, and sleep strategy of security cameras, sensors, and inspection modules. The core logic is that security energy consumption equals the comprehensive calculation result of device standby power consumption, operating power consumption, inspection duration, and sleep duration. The security response coverage rate in the optimization decision boundary is a constraint condition of the model, requiring the coverage rate to always remain at 100%. In the example, the standby power consumption of a single camera is 5 watts, the operating power consumption is 15 watts, the peak inspection frequency is once per hour, and the sleep strategy is activated during off-peak hours, with inspections every two hours. Increasing the standby time can reduce the total energy consumption. This model clearly represents the mapping relationship between inspection frequency, sleep strategy, and security energy consumption.

[0050] By integrating the energy consumption models of the three subsystems, the input parameters, constraints, and output energy consumption values ​​of each model are clearly defined, forming a standardized subsystem energy consumption mapping relationship. This relationship can directly convert the subsystem operating parameters into energy consumption values, achieving a precise correspondence between parameters and energy consumption, and providing a complete mapping basis for the subsequent construction of multi-objective optimization functions.

[0051] By utilizing the energy consumption mapping relationship of subsystems, a multi-objective optimization function is constructed with the goal of minimizing total energy consumption and constraints of thermal comfort, illuminance satisfaction, and security response coverage. Electricity price weighting factors are introduced to generate the optimization objective function. The core of this step is to build an optimization function based on the subsystem energy consumption mapping relationship, with the minimum total energy consumption as the core objective and indoor comfort and security as constraints. Time-of-use electricity pricing weighting factors are incorporated to achieve peak-valley differentiated optimization, ultimately generating a solvable optimization objective function. The specific implementation method is as follows: The core objective of the multi-objective optimization function is to minimize the total energy consumption of the building. The total energy consumption is the sum of lighting energy consumption, air conditioning energy consumption, and security energy consumption. By substituting the operating parameters of each subsystem into the calculation through the subsystem energy consumption mapping relationship, the objective expression of the function is that the total energy consumption equals the sum of lighting energy consumption, air conditioning energy consumption, and security energy consumption. All energy consumption units are unified in watts to ensure consistent calculation.

[0052] The constraints are set around the indoor user experience and system function assurance, and include three types of hard constraints. The first type is thermal comfort constraint, which uses the predicted average voting value (PMV) as the evaluation index, with a constraint range of -0.5 to +0.5. This range is the most comfortable range for human perception, and the air conditioning parameters must not be adjusted beyond this range. The second type is illuminance satisfaction constraint, which requires the actual illuminance value to reach more than 90% of the required illuminance to avoid affecting the visual experience due to excessive dimming. The third type is security response coverage constraint, which requires security equipment to have full coverage without blind spots, and the response coverage rate to always remain at 100%. All three types of constraints are insurmountable hard conditions to ensure that the optimization process does not reduce the quality of building use.

[0053] The electricity price weighting factor is dynamically introduced based on the time-of-use electricity price period. It is the core parameter for achieving peak-valley differentiated optimization. The factor value is proportional to the electricity price of the current period. The weighting factor is 1.2 during peak hours, 1.0 during normal hours, and 0.4 during valley hours. The weighting factor is multiplied by the energy consumption of each subsystem before being included in the total energy consumption, which amplifies the weight ratio of peak energy consumption and forces the optimization algorithm to prioritize reducing energy consumption during peak hours, thereby achieving the energy-saving effect of peak shaving and valley filling.

[0054] By integrating the core objectives, constraints, and electricity price weighting factors, a complete multi-objective optimization function is constructed. The function uses the operating parameters of each subsystem as optimization variables, minimizes total energy consumption as the optimization objective, uses three types of experience indicators as constraints, and uses electricity price weighting as an adjustment factor to form a standardized optimization objective function. This function can solve for the optimal operating parameters through intelligent algorithms to achieve a balance between energy consumption and experience.

[0055] Solve the objective function to obtain recommended operating parameters for each subsystem, including at least the air conditioning set temperature and start / stop time, lighting dimming curve, security inspection frequency and hibernation strategy, and generate the initial collaborative parameter set for the lighting, air conditioning and security subsystems.

[0056] The core of this step is to use intelligent optimization algorithms to solve the constructed multi-objective optimization function, obtain the optimal subsystem operating parameters that satisfy all constraints, integrate the parameters by time and region, and generate an initial cooperative parameter set to provide a foundation for subsequent thermal inertia compensation. The specific implementation method is as follows: The objective function is solved using a genetic algorithm, which is suitable for complex optimization problems with multiple constraints and variables. It has high solution accuracy and fast convergence speed. The algorithm is set with a population size of 50, 100 iterations, a crossover probability of 0.8, and a mutation probability of 0.1. Through iterative optimization generation by generation, the optimal parameter combination that minimizes total energy consumption and satisfies all constraints is found. The solution process takes the next 24 hours as the optimization period and calculates the optimal parameters hourly to ensure the temporal rationality of the parameters.

[0057] The obtained air conditioning subsystem parameters include set temperature, start / stop time, and supply air temperature. In the example, the office area air conditioning set temperature is 26 degrees Celsius during peak hours, which is 1 degree Celsius higher than the normal setting to reduce energy consumption. The start / stop time is 10 minutes earlier and 10 minutes later to avoid peak electricity prices. The normal set temperature is 25 degrees Celsius, and the off-peak temperature is 24 degrees Celsius to make full use of low-priced electricity.

[0058] The lighting subsystem parameters are hourly dimming curves, with the dimming rate dynamically adjusted between 0.2 and 1.0. In the example, the dimming rate is 0.5 in occupied areas during peak hours and drops to 0.2 in unoccupied areas. During off-peak hours, the dimming rate is appropriately increased according to the illuminance requirements. The dimming curve is linked to the outdoor light intensity, reducing the dimming rate on sunny days and increasing it on cloudy days to ensure that the illuminance satisfaction standard is met.

[0059] The security subsystem parameters include inspection frequency and sleep strategy. In the example, the inspection frequency is once per hour during peak hours, and a deep sleep strategy is enabled during off-peak hours, with the inspection frequency adjusted to once every two hours. The sampling interval of sensors in non-critical areas is extended to reduce energy consumption while ensuring 100% coverage.

[0060] All subsystem parameters obtained from the solution are integrated according to functional area and timestamp, and the parameter format and unit are unified. The applicable time period, applicable area and constraint conditions of each parameter are marked to form an initial collaborative parameter set containing hourly parameters for 24 hours. All parameters in the parameter set are matched with each other and there is no time conflict, which satisfies the optimization objective and constraint conditions, and can directly enter the next step of thermal inertia dynamic compensation.

[0061] S204, Perform dynamic compensation for building thermal inertia on the initial collaborative parameter set, and adjust the precooling and preheating start and stop times and supply air temperature setting curves of the air conditioning system according to the heat storage delay characteristics of the building envelope and the allowable fluctuation range of indoor thermal comfort, to generate a transition parameter set corrected by thermal inertia. Specifically, the material parameters of the building envelope for each functional area can be extracted from the building information model, including wall thickness, density, specific heat capacity and thermal conductivity, to generate a table of thermal property parameters of the building envelope. The core of this step is to accurately extract key physical parameters of the building envelope that affect thermal inertia from the standardized building information model, transforming the building's physical structure into quantitative data that can be used for thermal calculations, thus providing a basis for subsequent calculations of heat storage delay characteristics. The specific implementation method is as follows: Building Information Modeling (BIM) is a digital carrier that provides comprehensive physical information about a building. It includes complete structural information about the building envelope for each functional area, which is mainly divided into different types such as office areas, meeting rooms, corridors, computer rooms, and lobbies. The materials and dimensions of the building envelope vary between these functional areas, as do their thermal inertia characteristics. Therefore, it is necessary to extract parameters for each area to ensure calculation accuracy. The extraction process uses a real-time model interface to directly read the component attribute data of the building envelope from the model, eliminating the need for manual measurement. The extraction accuracy is controlled at the millimeter level and the precision of physical quantities is at the thousandths of a percent, ensuring that the parameters accurately reflect the actual thermal characteristics of the building.

[0062] The thermal properties of the building envelope are core indicators determining the heat storage and delay characteristics of a building. These include four key parameters: wall thickness, density, specific heat capacity, and thermal conductivity. Each parameter has a clear physical meaning and standard values. Wall thickness refers to the vertical thickness of the building envelope components such as external walls, internal partitions, floors, and roofs, measured in millimeters. In the example, the external wall thickness of the office area is 240 mm, the internal partition wall thickness is 120 mm, and the roof thickness is 300 mm. Thickness directly affects the path length for heat transfer. Density refers to the mass per unit volume of the building envelope material, measured in kilograms per cubic meter. The density of concrete walls is 2400 kg / m³, and the density of lightweight partition boards is 800 kg / m³. Higher density indicates stronger heat storage capacity. Specific heat capacity refers to the amount of heat required to raise the temperature of a unit mass of building envelope material by a unit, measured in kilojoules per kilogram of degree Celsius. The specific heat capacity of concrete is 0.84 kilojoules per kilogram of degree Celsius, while that of thermal insulation materials is 1.0 kilojoules per kilogram of degree Celsius. Specific heat capacity determines a material's ability to store heat. Thermal conductivity refers to a material's ability to transfer heat, measured in watts per meter of degree Celsius. A lower thermal conductivity indicates better insulation performance. The thermal conductivity of external wall insulation layers is 0.04 watts per meter of degree Celsius, while that of concrete walls is 1.74 watts per meter of degree Celsius. This parameter directly affects the rate of heat transfer.

[0063] The extraction process is conducted in functional zones, extracting four parameters for each functional zone's exterior walls, interior walls, roof, windows, and flooring components. These parameters are then organized according to a fixed format: functional zone name, component type, parameter name, parameter value, and unit. Duplicate and invalid data are removed. Missing parameters are supplemented using standard parameters for the same type of building. After supplementation, the parameters are validated for reasonableness to ensure they conform to the conventional range for building materials. All validated parameters are integrated into a structured thermal property parameter table for the building envelope. The table clearly marks all thermal property data for each functional zone. This parameter table can be directly imported into the thermal calculation module, providing standardized input data for subsequent finite difference method calculations.

[0064] Based on the thermal property parameter table of the building envelope, the one-dimensional unsteady heat conduction equation is solved by the finite difference method, the response delay time of indoor temperature to air conditioning start-up and shutdown is calculated, and the heat storage delay characteristic curves of each zone are generated. The core of this step is to solve the unsteady heat conduction process of the building envelope through numerical calculation methods, quantify the time delay between air conditioning start-up and shutdown and indoor temperature changes, and transform the abstract thermal inertia characteristics into a visualized time delay curve. The specific implementation method is as follows: The finite difference method is a commonly used numerical method for solving the differential equation of heat conduction. Its core logic involves discretizing the continuous heat-conducting region of the building envelope into a finite number of tiny nodes. A system of algebraic equations is established based on the heat transfer relationships between these nodes, and the temperature variation over time is obtained through iterative solutions. This method is suitable for calculating one-dimensional unsteady-state heat conduction in building envelopes. The calculation time step is set to 1 minute, and the spatial step is set to 10 millimeters, balancing computational accuracy and efficiency. The one-dimensional unsteady-state heat conduction equation is the core mathematical model describing the heat transfer within the building envelope over time. The equation includes parameters such as the partial derivative of temperature with respect to time, the second partial derivative of temperature with respect to space, thermal conductivity, density, and specific heat capacity. It comprehensively reflects the dynamic relationship between heat transfer, heat storage, and heat conduction, and is the theoretical basis for calculating thermal inertia delay.

[0065] Substituting the corresponding parameters from the thermal property parameter table of the building envelope into the one-dimensional unsteady-state heat conduction equation, calculation models were established for the main heat storage components such as the exterior walls and roof of each functional area. Boundary conditions such as outdoor temperature fluctuations, initial indoor temperature, and air conditioning supply air temperature were set to simulate the entire process of heat transfer from the air conditioning unit to the building envelope when it is turned on, and the release of heat stored in the building envelope to the interior when it is turned off. The change in indoor temperature at different times was obtained through iterative calculation. The time difference between the start and stop times of the air conditioning and the time when the indoor temperature reaches the set value is the response delay time, which is a direct reflection of the building's thermal inertia. In the example, due to the large thickness and high density of the exterior walls of the office area, the delay time for the indoor temperature to drop to the set value after the air conditioning is turned on is 15 minutes, and the delay time for the indoor temperature to rise again after the air conditioning is turned off is 22 minutes. The delay time of the lightweight partition walls in the conference room is shorter, with an 8-minute delay when turned on and a 12-minute delay when turned off. The roof has the strongest heat storage capacity, with a delay time of up to 30 minutes.

[0066] Using time as the horizontal axis and indoor temperature change or delay time as the vertical axis, the response delay data under different air conditioning operating conditions in each functional area are continuously plotted to form a smooth heat storage delay characteristic curve. Each curve corresponds to a functional area, clearly showing the rate of change of indoor temperature, delay time, and heat storage and release patterns after the air conditioner starts and stops. The curves include key features such as pre-cooling delay in cooling mode, pre-heating delay in heating mode, and temperature fluctuation delay after shutdown, which can intuitively reflect the differences in thermal inertia of each area and provide a quantitative basis for subsequent air conditioning parameter adjustments.

[0067] Based on the heat storage delay characteristic curve and the allowable fluctuation range of indoor thermal comfort, the start and stop times of the air conditioner in the initial coordination parameter set are adjusted in advance or in advance, and the supply air temperature setting curve is recalculated to generate a pre-cooling and preheating adjustment scheme. The core of this step is to combine thermal inertia delay characteristics with human thermal comfort constraints to optimize the start-up and stop times and supply air temperature of the air conditioner, and to optimize energy consumption by utilizing the heat storage characteristics of the building envelope, while ensuring that indoor comfort standards are met. The specific implementation method is as follows: The permissible fluctuation range for indoor thermal comfort is a temperature fluctuation range set based on human thermal comfort requirements. For summer cooling, this range is set to 24°C to 27°C, and for winter heating, it is set to 20°C to 24°C, with a temperature fluctuation not exceeding ±1.5°C. This range ensures occupant comfort while providing reasonable leeway for adjusting air conditioning parameters, serving as a constraint for thermal inertia compensation. The air conditioning start-up and stop times in the initial coordinated parameter set are based on direct calculations from load forecasts, without considering the heat storage delay of the building envelope. Direct execution of these parameters can lead to delayed indoor temperature fluctuations, resulting in excessive temperature or energy waste. Therefore, dynamic adjustments based on the heat storage delay characteristic curve are necessary.

[0068] The adjustment logic follows the principle of thermal inertia compensation. In cooling mode, the building's heat storage characteristics are utilized to pre-cool the air conditioner during off-peak electricity periods, storing the cold energy in the building envelope. The pre-cooling start time is advanced by one delay cycle according to the delay curve. In the example, the pre-cooling start time in the office area is advanced from 14:00 to 13:45, starting 15 minutes earlier, utilizing the building envelope to store the cold energy. During peak electricity periods, the air conditioner is turned off later, relying on the stored cold energy to maintain the indoor temperature. The shutdown time is delayed from 18:00 to 18:22, reducing energy consumption during peak periods. In heating mode, the air conditioner is started in advance for preheating and shut down later to utilize the stored heat for insulation, avoiding a rapid drop in temperature.

[0069] The air supply temperature setting curve is a continuous curve describing the change of air supply temperature over time. The original curve did not take into account the temperature lag caused by thermal inertia. During adjustment, it is recalculated in combination with the delay characteristics and load demand. In the cooling mode, the air supply temperature is appropriately reduced to 16 degrees Celsius during the pre-cooling stage to enhance the cold storage effect. During the normal operation stage, the air supply temperature is maintained at 18 degrees Celsius. Before shutdown, the air supply temperature is gradually increased to 20 degrees Celsius to smoothly transition and reduce temperature fluctuations. In the heating mode, the air supply temperature is set to 28 degrees Celsius during the preheating stage, 25 degrees Celsius during the normal operation stage, and reduced to 23 degrees Celsius before shutdown.

[0070] During the adjustment process, indoor temperature fluctuations are checked in real time to ensure that they are always within the allowable range of thermal comfort and to avoid a decrease in comfort due to excessive pre-cooling and pre-heating. The adjusted air conditioner start and stop times, pre-cooling and pre-heating durations, air supply temperatures at different times, and temperature control thresholds are integrated to form a pre-cooling and pre-heating adjustment scheme for different functional zones and time periods. The scheme includes two operating conditions, summer and winter, to adapt to the thermal inertia compensation needs of different seasons.

[0071] The precooling and preheating adjustment scheme is merged into the lighting and security parameters, the start and stop times and supply air temperature set values ​​of the air conditioning section are updated, and a set of transition parameters corrected by thermal inertia is generated.

[0072] The core of this step is to integrate the thermally inertia-corrected air conditioning parameters with the original lighting and security parameters, retain the rationality of non-air conditioning parameters, replace air conditioning-related parameters, and form a set of transitional parameters that take into account thermal inertia characteristics. This provides input for the subsequent resolution of subsystem coupling conflicts. The specific implementation method is as follows: The initial coordination parameter set includes the operating parameters of three subsystems: lighting, air conditioning, and security. The lighting parameters include the dimming curves of each functional area, the illuminance setpoint, and the start and stop times. The security parameters include the inspection frequency, camera sleep strategy, and intrusion detection sensitivity. The air conditioning parameters include the original start and stop times, supply air temperature, and operating mode. The pre-cooling and pre-heating adjustment scheme only makes thermal inertia corrections to the air conditioning parameters. The lighting and security parameters are not affected by thermal inertia and do not need to be adjusted; their original values ​​are directly retained.

[0073] The merging process adopts a parameter overwrite update method. First, the lighting subsystem parameters and security subsystem parameters in the initial collaborative parameter set are extracted, keeping the parameter format, value, and timing completely unchanged to ensure that the operation logic of the two subsystems is not affected. Then, the air conditioning correction parameters in the pre-cooling and preheating adjustment scheme are extracted, including the adjusted pre-cooling and preheating start and stop times, the air supply temperature setting curves for each time period, and the temperature fluctuation constraint values. The corrected air conditioning parameters completely replace the original air conditioning parameters in the initial parameter set to achieve accurate updating of air conditioning parameters.

[0074] After the update, the integrated parameters were time-series verified to ensure that the timestamps of all parameters were consistent, that there were no time-series conflicts between the air conditioning start / stop times and the operating periods of lighting and security, that the supply air temperature setting curve matched the load period and electricity price period, and that the parameters of each functional area corresponded independently without overlap or confusion. In the example, the office area lighting dimming curve maintained an automatic adjustment from 100 lux to 300 lux, the security inspection frequency remained once every 30 minutes, the air conditioning start time was updated to 13:45, and the supply air temperature curve was reset according to the three stages of pre-cooling, operation, and shutdown.

[0075] The verified integrated parameters are categorized by subsystem and arranged chronologically to form a standardized set of transitional parameters corrected for thermal inertia. The parameter set includes complete operating parameters for lighting, air conditioning, and security. Among them, the air conditioning parameters have undergone dynamic compensation for thermal inertia, taking into account both energy consumption optimization and thermal comfort requirements. This parameter set serves as an intermediate output result, providing stable and accurate input data for the next step of subsystem coupling analysis and conflict resolution.

[0076] S205, perform inter-system coupling analysis and conflict resolution on the transition parameter set, adjust the contradictory operating parameters through a fuzzy priority arbitration mechanism, output a set of conflict-free subsystem collaborative operating parameters, and send them to the lighting controller, air conditioning controller and security controller for execution, so as to realize the adaptive energy consumption optimization control of the intelligent building system.

[0077] Specifically, it can analyze the operating parameters of the three subsystems of centralized lighting, air conditioning, and security, identify the coupling relationship between parameters, and generate a coupling influence factor matrix; The core of this step is to decompose the transition parameter set and perform subsystem correlation analysis, accurately extract the operating parameters of each subsystem and identify the coupling relationships between them, and construct a standardized matrix by quantifying the degree of influence to provide a data foundation for subsequent conflict detection. The specific implementation method is as follows: The transition parameter set is a collection of multi-system operating parameters after dynamic compensation for building thermal inertia. It includes all control parameters for the three core subsystems: lighting, air conditioning, and security. Each parameter is accompanied by a timestamp and area identifier, covering all functional areas within the building, including offices, meeting rooms, corridors, and computer rooms. It serves as the core basis for collaborative control. The analysis process employs a time-series parameter analysis module, classifying and extracting parameters according to subsystem type. It breaks down each parameter's value, duration of action, control area, and execution logic to ensure no parameter is omitted.

[0078] The operating parameters of the lighting subsystem mainly include the dimming rate of each area, the on / off control period, the color temperature adjustment value, and the sensor trigger threshold. The dimming rate ranges from 0 to 100%, with higher values ​​indicating higher brightness. The on / off control period is synchronized with the building's operating hours, and the sensor trigger threshold is used to determine the brightness activation conditions when people are present. The operating parameters of the air conditioning subsystem include the pre-cooling and pre-heating start / stop times, the supply air temperature setpoint, the fan speed level, and the area temperature control range. The supply air temperature setpoint covers 16 to 30 degrees Celsius, and the fan speed levels are divided into low, medium, and high, corresponding to different air circulation efficiencies. The operating parameters of the security subsystem include the inspection frequency, the camera rotation angle, the device sleep period, and the intrusion response sensitivity. The inspection frequency is measured in hours, and the sleep period is set during nighttime when there is no activity to reduce ineffective energy consumption.

[0079] The coupling relationship refers to the associated characteristics of mutual influence and mutual restriction among the operating parameters of different subsystems. It is the key analysis object for the coordinated control of multiple systems in intelligent buildings. In the identification process, combined with the physical characteristics of the building and the operating logic of equipment, three core coupling relationships are comprehensively sorted out. The first type is the thermal coupling between air conditioning and lighting. The lighting fixtures generate heat when emitting light. The cooling load of the air conditioner is positively correlated with the lighting brightness. When the lighting dimming rate increases, the cooling load demand of the air conditioner will increase, and the supply air temperature needs to be adjusted synchronously. The second type is the spatial coupling between air conditioning and security. When the air conditioner precools or preheats, it requires the building area to be kept closed, while the security inspection needs to open the area door. The conflict in the operating time periods of the two will affect the temperature control effect and security. The third type is the linkage coupling between lighting and security. When the security inspection starts, the lighting brightness in the area needs to be increased, while the lighting energy-saving strategy requires reducing the brightness. The conflict in parameter settings will lead to insufficient inspection vision or energy waste.

[0080] The coupling influence factor is a numerical index that quantifies the strength of the coupling relationship. The value range is from 0 to 1. The larger the value, the more significant the coupling influence between the two parameters. No influence is 0, and strong coupling is 1. Based on the identified coupling relationship, a corresponding coupling influence factor is assigned to each group of associated parameters. For example, the coupling influence factor between the lighting dimming rate and the air conditioner cooling load is 0.85, the coupling influence factor between the air conditioner precooling period and the security inspection time is 0.9, and the coupling influence factor between the lighting brightness and the security inspection vision is 0.75. Taking all subsystem parameters as the matrix row and column dimensions, the coupling influence factor values are filled in the corresponding positions to generate a coupling influence factor matrix with regular rows and columns. The matrix fully presents the coupling strength between all parameters, providing an accurate quantitative basis for subsequent conflict detection.

[0081] Detect parameter combinations that conflict with each other according to the coupling influence factor matrix, mark the conflict parameter pairs and calculate the conflict degree coefficient, and generate a list of conflict parameters; The core of this step is to rely on the coupling influence factor matrix to locate the parameter conflict points, quantify the severity of the conflict through a mathematical model, mark the conflict combinations according to the priority, and form a list of conflict parameters that can be executed for correction. The specific implementation method is as follows: Parameter conflict means that two or more subsystem parameters with coupling associations generate mutually contradictory operation instructions when executed in the same time period and the same area, resulting in increased energy consumption, decreased comfort, or security failure. It is the core problem that needs to be solved in the coordinated control of intelligent buildings. The detection process is based on the coupling influence factor matrix. Set the coupling influence factor greater than 0.7 as the strong association threshold, and screen out the parameter combinations in the matrix whose values exceed the threshold as the potential conflict candidate set. This threshold can adapt to the equipment operation characteristics of most public buildings, taking into account both the detection sensitivity and accuracy.

[0082] Logical verification is performed on parameter combinations in the potential conflict candidate set to determine whether there are contradictions in the parameter execution instructions. For example, the air conditioning in the conference room area is set to start pre-cooling at 2 PM, requiring the area to be closed, while the security parameters are set to perform inspections in the area at 2 PM, requiring the access control to be opened. These two instructions directly conflict. The lighting parameters in the office area are set to have a dimming rate of 30% for energy saving from 1 PM to 3 PM, while the security parameters are set to inspect every half hour during this period. The low brightness will cause the inspection screen to be blurry, which is an indirect conflict. All conflicting parameter pairs that pass verification are marked, clearly indicating the name of the conflicting parameter, its subsystem, its area of ​​effect, and the execution time period, ensuring that the conflict location is clearly traceable.

[0083] The conflict severity coefficient is a core indicator for quantifying the severity of conflict. It ranges from 0 to 1 and is calculated by multiplying the coupling influence factor, parameter execution deviation rate, and regional importance coefficient. The coupling influence factor reflects the strength of the association, the parameter execution deviation rate reflects the degree of instruction contradiction, and the regional importance coefficient reflects the scope of the conflict's impact. Core areas such as office areas and server rooms have a coefficient of 1, while secondary areas such as corridors and lounges have a coefficient of 0.6. In the example, the conflicting parameter pair between conference room air conditioning pre-cooling and security patrol has a coupling influence factor of 0.9, a parameter execution deviation rate of 1, a regional importance coefficient of 1, and a conflict severity coefficient of 0.9, classifying it as a severe conflict. The conflicting parameter pair between office area lighting dimming and security patrol has a coupling influence factor of 0.75, a parameter execution deviation rate of 0.8, a regional importance coefficient of 1, and a conflict severity coefficient of 0.6, classifying it as a moderate conflict.

[0084] All conflicting parameter pairs are sorted from highest to lowest according to their conflict severity coefficients. The parameter name, its subsystem, conflict area, conflict period, conflict severity coefficient, and conflict cause are recorded in sequence to form a standardized conflicting parameter list. The list visually presents the severity and distribution of all conflicts, providing a clear correction target for subsequent fuzzy priority arbitration and parameter adjustment.

[0085] Based on the fuzzy priority arbitration mechanism, priority weights are assigned to each subsystem according to time periods, and conflicting parameters are adjusted by compromise according to weights to generate a set of parameters for the coordinated operation of non-conflicting subsystems. The core of this step is to balance the control objectives of each subsystem through a fuzzy priority arbitration mechanism, dynamically allocate weights according to the building's operating time, rationally adjust conflicting parameters, eliminate parameter contradictions, and generate collaborative operating parameters that are suitable for all time periods. The specific implementation method is as follows: The fuzzy priority arbitration mechanism is an intelligent decision-making method to resolve parameter conflicts among multiple subsystems. Its core logic is to assign differentiated priority weights to the lighting, air conditioning, and security subsystems based on the core needs of different operating periods of the building. The weight values ​​range from 0 to 1, and the sum of the weights of the three subsystems is always 1. The higher the priority weight, the higher the priority of the control objective of the subsystem during this period. The mechanism can dynamically adapt to the building's operating schedule and avoid the control rigidity caused by fixed priorities.

[0086] Based on the typical operating periods of intelligent buildings, five scenarios are divided into weekday working hours, lunch break hours, after-get off work hours, nighttime hours, and holiday hours, and subsystem priority weights are assigned to each scenario. During weekday working hours from 8:00 to 18:00, the core building requirements are indoor thermal comfort and lighting comfort, with air conditioning having a priority weight of 0.45, lighting a priority weight of 0.4, and security a priority weight of 0.15. During lunch break hours from 12:00 to 14:00, the focus is on low energy consumption and a quiet environment, with lighting a priority weight of 0.35, air conditioning a priority weight of 0.4, and security a priority weight of 0.25. During the evening hours from 18:00 to 22:00, energy conservation is the primary concern, with lighting a priority weight of 0.2, air conditioning a priority weight of 0.3, and security a priority weight of 0.5. During the night hours from 22:00 to 8:00 the next day, the core requirement is security, with security a priority weight of 0.7, air conditioning a priority weight of 0.2, and lighting a priority weight of 0.1. During holidays when there is no human activity, security has a priority weight of 0.6, and both air conditioning and lighting have a priority weight of 0.2.

[0087] For each set of conflicting parameters in the conflict parameter list, a compromise adjustment is made based on the priority weight of the corresponding time period. The adjustment principle is to prioritize satisfying the parameter settings of high-priority subsystems, while making appropriate compromises to low-priority subsystem parameters within permissible limits, ensuring that the adjusted parameters do not have execution conflicts, while also taking into account energy consumption optimization and functional requirements. In the example, there is a conflict between the conference room air conditioning pre-cooling and security inspection during working hours, with the air conditioning priority (0.45) being higher than the security priority (0.15). The adjustment plan is to postpone the security inspection time by 15 minutes to avoid the air conditioning pre-cooling shutdown period, while keeping the air conditioning pre-cooling parameters unchanged. There is also a conflict between office area lighting dimming and security inspection, with the lighting priority (0.4) being higher than the security priority (0.15). The adjustment plan is to increase the lighting dimming rate to 50% during the inspection period, which satisfies the security visibility requirements without completely abandoning the energy-saving goal.

[0088] After all conflicting parameters are adjusted, the corrected air conditioning, lighting, and security parameters are integrated with the original parameters that do not conflict. Duplicate parameters are removed, time period logic is calibrated, and area identifiers are unified to generate a set of subsystem collaborative operation parameters that covers the entire building, all time periods, and has no execution conflicts. The parameter set can be directly sent to the field controller for execution to ensure the stable collaborative operation of multiple systems.

[0089] The set of parameters for the coordinated operation of conflict-free subsystems is distributed to the lighting controller, air conditioning controller and security controller for execution, and the execution status is returned to realize the adaptive energy consumption optimization control of the intelligent building system.

[0090] The core of this step is to complete the distributed distribution of collaborative parameters and the closed-loop feedback of execution status. Control commands are implemented through field controllers, and the operating effect is monitored in real time to achieve closed-loop control for adaptive energy consumption optimization. The specific implementation method is as follows: The parameter set for conflict-free subsystem collaborative operation is transmitted via industrial Ethernet. The communication protocol is compatible with the building intelligent control system standard, and the transmission delay is controlled within 100 milliseconds to ensure that parameter commands are delivered to the controller in real time. The transmission process is categorized by subsystem type and control area. Lighting controllers receive lighting parameters such as dimming rate, on / off time, and color temperature adjustment; air conditioning controllers receive air conditioning parameters such as pre-cooling / preheating start / stop time, air supply temperature, and fan speed; and security controllers receive security parameters such as inspection frequency, sleep time, and response sensitivity. Before transmission, the parameters are verified to ensure that the values ​​are within the allowable operating range of the equipment and to avoid invalid commands damaging the equipment.

[0091] The lighting, air conditioning, and security controllers are all distributed field control terminals. Upon receiving parameter commands, they immediately parse and execute them. The lighting controller adjusts the brightness of the lamps based on dimming rates and time-of-use parameters to achieve on-demand lighting. The air conditioning controller operates according to start / stop times and supply air temperature curves, combining building thermal inertia for precise temperature control and reduced energy consumption. The security controller operates according to inspection frequency and sleep strategies, ensuring safety while reducing standby power consumption. During execution, the controllers collect real-time equipment operating status data, including actual lighting illuminance, air conditioning supply air temperature and indoor temperature of the area, security inspection completion status, and equipment operating power. Data is collected every minute to ensure real-time status monitoring.

[0092] The execution status feedback adopts a two-way communication mode. The controller synchronously uploads the collected operating data and command execution results to the central control platform. The feedback data includes information such as area identifier, equipment number, execution parameters, real-time monitoring values, and operation fault codes. The central control platform analyzes the feedback data to determine whether the parameters are executed correctly, whether the operating effect meets the standard, and whether there are any equipment abnormalities. If an execution deviation occurs, such as the air conditioning supply temperature not reaching the set value or the lighting dimming rate deviation being too large, the platform immediately activates the adaptive correction logic, fine-tunes the corresponding parameters, and reissues them, forming a closed-loop control of "issuance-execution-feedback-correction".

[0093] Through closed-loop collaborative control, the lighting, air conditioning, and security subsystems operate in a conflict-free manner. While meeting indoor thermal comfort, illuminance requirements, and security needs, the system minimizes the building's total energy consumption. It responds in real time to changes in occupancy density, meteorological parameters, and electricity prices, continuously optimizing operating parameters to ultimately achieve adaptive energy consumption optimization control of the intelligent building system, thus achieving the dual goals of energy saving and comfort.

[0094] Another embodiment of the present invention provides an adaptive energy consumption optimization control system for an intelligent building system, see [link to relevant documentation]. Figure 3 The system may include: The data acquisition module 301 is used to collect real-time data on personnel density, outdoor meteorological parameters and real-time electricity price information in various areas of the building, and generate a multi-source dynamic environmental feature set. Prediction module 302 is used to calculate the hourly cooling and heating load and illuminance demand of each functional area within a future preset time period based on the multi-source dynamic environment feature set and through a machine learning prediction model, and output a regional load demand distribution map. The construction module 303 is used to construct a multi-objective optimization function with the goal of minimizing total energy consumption based on the regional load demand distribution map and the real-time electricity price time information, and to generate an initial set of collaborative parameters for lighting, air conditioning and security subsystems. The adjustment module 304 is used to perform dynamic compensation for building thermal inertia on the initial collaborative parameter set. Based on the heat storage delay characteristics of the building envelope and the allowable fluctuation range of indoor thermal comfort, the module adjusts the precooling and preheating start and stop times and the supply air temperature setting curve of the air conditioning system to generate a transition parameter set corrected by thermal inertia. The control module 305 is used to perform inter-subsystem coupling analysis and conflict resolution on the transition parameter set, adjust the contradictory operating parameters through a fuzzy priority arbitration mechanism, output a set of conflict-free subsystem collaborative operating parameters, and send them to the lighting controller, air conditioning controller and security controller for execution, so as to realize the adaptive energy consumption optimization control of the intelligent building system.

[0095] This invention also provides a storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above method embodiments when running.

[0096] This invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0097] Specifically, the aforementioned electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the aforementioned processor, and the input / output device is connected to the aforementioned processor.

[0098] The above description, based on the embodiments shown in the figures, details the structure, features, and effects of the present invention. The above description is only a preferred embodiment of the present invention, but the present invention is not limited to the scope of implementation shown in the figures. Any changes made in accordance with the concept of the present invention, or equivalent embodiments modified to have equivalent changes, that do not exceed the spirit covered by the specification and figures, should be within the protection scope of the present invention.

Claims

1. An adaptive energy consumption optimization control method for an intelligent building system, characterized in that, The method includes: Real-time data collection of personnel density, outdoor meteorological parameters, and real-time electricity price information in various areas of the building is used to generate a multi-source dynamic environmental feature set. Based on the multi-source dynamic environmental feature set, the hourly cooling and heating loads and illuminance requirements of each functional area within a preset time period are calculated using a machine learning prediction model, and a regional load demand distribution map is output. Based on the regional load demand distribution map and the real-time electricity price time information, a multi-objective optimization function with the goal of minimizing total energy consumption is constructed to generate the initial collaborative parameter set for lighting, air conditioning and security subsystems. The initial collaborative parameter set is dynamically compensated for by building thermal inertia. Based on the heat storage delay characteristics of the building envelope and the allowable fluctuation range of indoor thermal comfort, the precooling and preheating start and stop times and supply air temperature setting curves of the air conditioning system are adjusted to generate a transition parameter set corrected by thermal inertia. The set of transition parameters is subjected to inter-system coupling analysis and conflict resolution. The conflicting operating parameters are adjusted through a fuzzy priority arbitration mechanism, and a set of conflict-free subsystem collaborative operating parameters is output and sent to the lighting controller, air conditioning controller and security controller for execution, so as to realize the adaptive energy consumption optimization control of the intelligent building system.

2. The method according to claim 1, characterized in that, The system collects real-time data on population density in various areas of the building, outdoor meteorological parameters, and real-time electricity price information to generate a multi-source dynamic environmental feature set, including: Real-time statistics on the number of people entering and exiting each functional area and calculation of instantaneous personnel density are generated to produce time-series data on regional personnel density. Collect outdoor temperature, humidity, solar radiation intensity and wind speed data, and simultaneously obtain real-time electricity price information to generate raw outdoor weather and electricity price datasets; The time-series data of regional population density and the raw data of outdoor weather and electricity prices are time-stamped and aligned. The missing sampling points are filled in by linear interpolation to generate a time-synchronized multi-source data matrix. The time-synchronized multi-source data matrix is ​​normalized and feature-encoded, and the electricity price period is converted into a tiered numerical label, ultimately generating a multi-source dynamic environment feature set.

3. The method according to claim 2, characterized in that, Based on the multi-source dynamic environmental feature set, the hourly cooling and heating loads and illuminance requirements of each functional area within a preset future time period are calculated using a machine learning prediction model, and a regional load demand distribution map is output, including: The multi-source dynamic environment feature set is divided into an input sliding window sequence according to the time order, and the feature data from one hour before the current time to the current time are extracted to generate the prediction input tensor. The prediction input tensor is input into a pre-trained long short-term memory network model. This model has learned the mapping relationship between historical load data and environmental characteristics, and outputs the predicted cooling load and heating load of each functional area in the future preset time period, generating a cooling and heating load prediction matrix. A multilayer perceptron model is used to process the nonlinear relationship between personnel density and illuminance demand. Combined with outdoor light intensity data, the illuminance demand values ​​of each functional area in the future preset time period are predicted, and an illuminance demand prediction vector is generated. The heating and cooling load prediction matrix and the illuminance demand prediction vector are spatially overlaid and visualized to generate a color heat map that identifies the hourly load values ​​of each functional area, and finally outputs a regional load demand distribution map.

4. The method according to claim 3, characterized in that, The step involves constructing a multi-objective optimization function with the goal of minimizing total energy consumption based on the regional load demand distribution map and the real-time electricity price time information, and generating an initial set of coordinated parameters for the lighting, air conditioning, and security subsystems, including: Analyze the hourly heating and cooling loads and illuminance demand in the load demand distribution map, combine the peak and valley periods of electricity price to determine the energy cost sensitivity of each period, and generate an optimization decision boundary with priority weights. Based on the optimized decision boundary, lighting energy consumption models, air conditioning energy consumption models, and security energy consumption models are established respectively, and subsystem energy consumption mapping relationships are generated. By utilizing the energy consumption mapping relationship of subsystems, a multi-objective optimization function is constructed with the goal of minimizing total energy consumption and constraints of thermal comfort, illuminance satisfaction, and security response coverage. Electricity price weighting factors are introduced to generate the optimization objective function. Solve the objective function to obtain recommended operating parameters for each subsystem, including at least the air conditioning set temperature and start / stop time, lighting dimming curve, security inspection frequency and hibernation strategy, and generate the initial collaborative parameter set for the lighting, air conditioning and security subsystems.

5. The method according to claim 4, characterized in that, The process of performing dynamic compensation for building thermal inertia on the initial collaborative parameter set involves adjusting the precooling and preheating start / stop times and supply air temperature set curves of the air conditioning system based on the heat storage delay characteristics of the building envelope and the allowable fluctuation range of indoor thermal comfort, thereby generating a transition parameter set corrected for thermal inertia. This includes: Extract the building envelope material parameters of each functional area from the building information model, including wall thickness, density, specific heat capacity and thermal conductivity, and generate a table of thermal property parameters of the building envelope. Based on the thermal property parameter table of the building envelope, the one-dimensional unsteady heat conduction equation is solved by the finite difference method, the response delay time of indoor temperature to air conditioning start-up and shutdown is calculated, and the heat storage delay characteristic curves of each zone are generated. Based on the heat storage delay characteristic curve and the allowable fluctuation range of indoor thermal comfort, the start and stop times of the air conditioner in the initial coordination parameter set are adjusted in advance or in advance, and the supply air temperature setting curve is recalculated to generate a pre-cooling and preheating adjustment scheme. The precooling and preheating adjustment scheme is merged into the lighting and security parameters, the start and stop times and supply air temperature set values ​​of the air conditioning section are updated, and a set of transition parameters corrected by thermal inertia is generated.

6. The method according to claim 5, characterized in that, The process involves performing inter-system coupling analysis and conflict resolution on the transition parameter set, adjusting conflicting operating parameters through a fuzzy priority arbitration mechanism, outputting a conflict-free subsystem collaborative operating parameter set, and distributing it to the lighting controller, air conditioning controller, and security controller for execution. This achieves adaptive energy consumption optimization control of the intelligent building system, including: The operation parameters of the three subsystems of centralized lighting, air conditioning and security are analyzed in the transition parameters, the coupling relationship between the parameters is identified, and the coupling influence factor matrix is ​​generated. Conflicting parameter combinations are detected based on the coupling influence factor matrix, conflicting parameter pairs are marked and conflict degree coefficients are calculated to generate a list of conflicting parameters. Based on the fuzzy priority arbitration mechanism, priority weights are assigned to each subsystem according to time periods, and conflicting parameters are adjusted by compromise according to weights to generate a set of parameters for the coordinated operation of non-conflicting subsystems. The set of parameters for the coordinated operation of conflict-free subsystems is distributed to the lighting controller, air conditioning controller and security controller for execution, and the execution status is returned to realize the adaptive energy consumption optimization control of the intelligent building system.

7. An adaptive energy consumption optimization control system for an intelligent building system, characterized in that, The system includes: The data acquisition module is used to collect real-time data on personnel density, outdoor meteorological parameters, and real-time electricity price information in various areas of the building, and generate a multi-source dynamic environmental feature set. The prediction module is used to calculate the hourly cooling and heating loads and illuminance requirements of each functional area within a preset time period based on the multi-source dynamic environmental feature set and through a machine learning prediction model, and output a regional load demand distribution map. The construction module is used to construct a multi-objective optimization function with the goal of minimizing total energy consumption based on the regional load demand distribution map and the real-time electricity price time information, and to generate the initial collaborative parameter set of lighting, air conditioning and security subsystems; The adjustment module is used to perform dynamic compensation for building thermal inertia on the initial collaborative parameter set. Based on the heat storage delay characteristics of the building envelope and the allowable fluctuation range of indoor thermal comfort, the module adjusts the precooling and preheating start and stop times and supply air temperature setting curves of the air conditioning system to generate a transition parameter set corrected for thermal inertia. The control module is used to perform inter-subsystem coupling analysis and conflict resolution on the transition parameter set, adjust the contradictory operating parameters through a fuzzy priority arbitration mechanism, output a set of conflict-free subsystem collaborative operating parameters, and send them to the lighting controller, air conditioning controller and security controller for execution, so as to realize the adaptive energy consumption optimization control of the intelligent building system.

8. The system according to claim 7, characterized in that, The acquisition module is specifically used for: Real-time statistics on the number of people entering and exiting each functional area and calculation of instantaneous personnel density are generated to produce time-series data on regional personnel density. Collect outdoor temperature, humidity, solar radiation intensity and wind speed data, and simultaneously obtain real-time electricity price information to generate raw outdoor weather and electricity price datasets; The time-series data of regional population density and the raw data of outdoor weather and electricity prices are time-stamped and aligned. The missing sampling points are filled in by linear interpolation to generate a time-synchronized multi-source data matrix. The time-synchronized multi-source data matrix is ​​normalized and feature-encoded, and the electricity price period is converted into a tiered numerical label, ultimately generating a multi-source dynamic environment feature set.

9. A storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute the method of any one of claims 1-6 when it is run.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to perform the method of any one of claims 1-6.

Images