Building equipment intelligent control optimization management system based on green low-carbon building

Abstract

The invention relates to the technical field of building intelligent control and energy saving, and discloses a building equipment intelligent control optimization management system based on a green low-carbon building. The system comprises five subsystems, wherein an environmental parameter sensing subsystem is used for acquiring temperature and humidity gradients, illumination intensity distribution and personnel activity thermodynamic diagram data inside and outside a building through a distributed sensing network to form a multi-dimensional environmental state vector; the equipment operation state acquisition subsystem acquires an operation power curve, an energy efficiency conversion rate curve and an equipment health degree index of heating ventilation air conditioning, illumination and renewable energy equipment in real time; the dynamic load prediction subsystem establishes an energy consumption load space-time distribution prediction model according to historical data time sequence relevance; the control strategy generation subsystem is used for generating an optimized instruction set containing an equipment start-stop time sequence, a power regulation gradient and an energy distribution weight in combination with the current environment vector and a predicted load; the execution terminal adapts the subsystem to convert the instruction into a device compatible signal and calibrate a response delay.

Description

Technical Field

[0001] This invention relates to the field of building intelligent control and energy-saving technology, specifically to an intelligent control and optimization management system for building equipment based on green and low-carbon buildings. Background Technology

[0002] Under the global trend of low-carbon development, the construction industry, as a key sector of energy consumption and carbon emissions, is facing an increasingly urgent need for green transformation. Traditional building equipment management models, limited by extensive control logic, are unable to meet the requirements of refined operation in green buildings.

[0003] At the environmental perception level, existing systems mostly rely on parameter collection from single or localized points, lacking the full-area coverage capability of distributed sensor networks. They can only acquire single temperature, humidity, or light intensity values, failing to generate multi-dimensional environmental data that includes temperature and humidity gradients, light intensity distribution, and heat maps of human activity. This one-sidedness in data collection makes it difficult for the system to accurately capture environmental differences in different areas and at different times within a building. For example, changes in personnel density in office areas and corridors of commercial complexes, as well as temperature stratification on different floors, cannot be effectively perceived.

[0004] In equipment management, traditional solutions lack sufficient dimensions for collecting operational data, focusing primarily on the basic start-up and shutdown status of conventional equipment such as HVAC and lighting. They lack real-time tracking of operating power curves and energy efficiency conversion rate curves, and fail to incorporate the operational status of renewable energy equipment into a unified monitoring system. Furthermore, the absence of equipment health indicators means that maintenance is often conducted only after a failure, failing to detect potential equipment anomalies in advance and frequently resulting in energy waste due to inefficient equipment operation.

[0005] In terms of load forecasting and control strategies, existing technologies mostly rely on simple predictions based on fixed time periods or single parameters, failing to establish time-series correlation models between historical environmental conditions and equipment operating data. This results in load forecasts lacking spatiotemporal distribution characteristics and being unable to predict changes in energy demand in different regions in advance. The corresponding control commands are mostly single start / stop or fixed power adjustments, neglecting the coordinated operation between equipment and lacking dynamic allocation mechanisms for renewable and traditional energy sources. For example, during periods of abundant sunshine, priority is not given to allocating photovoltaic power to lighting equipment, leading to insufficient utilization of clean energy.

[0006] At the execution level, building equipment from different brands and periods uses differentiated protocol stacks, and existing control systems lack universal adaptability, resulting in ineffective command transmission across devices. Simultaneously, the response delay of actuators is not taken into account in control measures, often leading to a disconnect between command issuance and actual execution, further reducing control accuracy. These problems combine to keep building equipment operating at consistently low energy efficiency, hindering the effective implementation of green and low-carbon buildings. Summary of the Invention

[0007] The purpose of this invention is to provide an intelligent control and optimization management system for building equipment based on green and low-carbon buildings, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides an intelligent control and optimization management system for building equipment based on green and low-carbon buildings, the system comprising:

[0009] The environmental parameter sensing subsystem collects temperature and humidity gradient data, light intensity distribution data, and human activity heat map data inside and outside the building through a distributed sensor network to form a multi-dimensional environmental state vector.

[0010] The equipment operation status acquisition subsystem acquires real-time operating power curves, energy efficiency conversion rate curves, and equipment health indicators for HVAC units, lighting equipment groups, and renewable energy equipment.

[0011] The dynamic load forecasting subsystem establishes a spatiotemporal distribution forecasting model for building energy load based on the temporal correlation between historical environmental state vectors and equipment operating status data.

[0012] The control strategy generation subsystem combines the current environmental state vector with the load distribution data output by the prediction model to generate an optimized control instruction set that includes equipment start-up and shutdown timing, power regulation gradient, and energy allocation weights.

[0013] The execution terminal adaptation subsystem converts the optimized control instruction set into control signals compatible with different building equipment protocol stacks and dynamically calibrates the response delay parameters of the actuators.

[0014] Preferably, the generation process of the multidimensional environment state vector is as follows:

[0015] The temperature and humidity gradient data are hierarchically clustered according to the building spatial topology to form a temperature and humidity field matrix with spatial continuity;

[0016] The light intensity distribution data is convolved with the light transmittance characteristics of the building facade to generate the corrected indoor light distribution tensor.

[0017] Based on the movement trajectory characteristics of personnel activity heatmap data, extract the coordinate set of high-frequency activity areas and the stay duration sequence;

[0018] After aligning the temperature and humidity field matrix, the modified illumination distribution tensor, and the personnel coordinate set in time and space, a multidimensional environmental state vector is constructed.

[0019] Preferably, the process for obtaining the device health index is as follows:

[0020] The vibration spectrum characteristics of the HVAC unit compressor, the refrigerant pressure fluctuation curve, and the fouling coefficient of the heat exchanger surface were collected.

[0021] Monitor the voltage harmonic distortion rate, light source attenuation index, and circuit insulation impedance changes of lighting equipment groups;

[0022] Analyze the cleanliness degradation curve of photovoltaic panels, the wear of wind turbine bearings, and the efficiency offset of inverters in renewable energy equipment.

[0023] By combining the characteristic data of the three types of equipment, a standardized equipment health index is output through a degradation status assessment model.

[0024] Preferably, the process of establishing the spatiotemporal distribution prediction model is as follows:

[0025] Extract the timestamps of temperature and humidity abrupt events and the light intensity transition intervals from the historical environmental state vector;

[0026] Match the peak power fluctuations and energy efficiency inflection points in the equipment's operating status data;

[0027] Establish a nonlinear mapping relationship between the rate of change of environmental parameters and the intensity of equipment load response;

[0028] A spatiotemporal distribution prediction model considering building thermal inertia is constructed using a long short-term memory network.

[0029] Preferably, the process of generating the optimized control instruction set is as follows:

[0030] Identify the deviation of abnormal parameters in the current environmental state vector and their spatial influence range;

[0031] Calculate the safety margin ratio between the predicted load distribution data and the rated capacity of the equipment;

[0032] The priority weight coefficients of each device are dynamically adjusted based on the device health index.

[0033] An optimized control instruction set containing multi-objective constraints is generated by combining the safety margin ratio and weighting coefficients.

[0034] Preferably, the adjustment process of the priority weight coefficient is as follows:

[0035] When the health index of the HVAC unit is lower than the first threshold, reduce its power regulation gradient upper limit.

[0036] When the health index of a group of lighting devices falls below the second threshold, the rate of change of its dimming frequency is limited.

[0037] When the health index of renewable energy equipment falls below the third threshold, its energy allocation weight compensation coefficient is increased.

[0038] Preferably, the process of converting the optimized control instruction set into control signals compatible with different building equipment protocol stacks is as follows:

[0039] Analyze and optimize the device operation codes and parameter adjustment values ​​in the control instruction set;

[0040] Match the communication protocol stack field structure and data encapsulation format of the target device;

[0041] While converting the opcode into a device-specific instruction set, time stamp synchronization information is inserted;

[0042] The transmission time offset of the control signal is pre-compensated based on the response delay parameter.

[0043] Preferably, the process for calibrating the response delay parameter of the dynamic calibration actuator is as follows:

[0044] Record the time difference between the control signal transmission time and the device status feedback reception time;

[0045] Analyze the latency fluctuation characteristics of different equipment types under different load rates;

[0046] Establish a dynamic correlation model between device response latency and real-time load rate;

[0047] Update the response latency parameters based on the current load rate forecast.

[0048] Preferably, the system further includes:

[0049] The energy efficiency assessment subsystem calculates the ratio of energy consumption intensity per unit area to carbon emission intensity for each equipment group in real time.

[0050] The calculation results were compared item by item with the thresholds of the green building certification standard.

[0051] The energy efficiency deviation correction coefficient is generated and fed back to the control strategy generation subsystem.

[0052] Preferably, the generation process of the energy efficiency deviation correction coefficient is as follows:

[0053] Identify the types of equipment whose energy consumption intensity per unit area exceeds the benchmark value and their spatial distribution;

[0054] Analyze the equipment operation combination patterns in areas with abnormal carbon emission intensity ratios;

[0055] Eliminate non-fault performance deviations based on equipment health indicators;

[0056] Generate energy efficiency deviation correction factors for specific equipment combinations.

[0057] Compared with the prior art, the beneficial effects of the present invention are:

[0058] The environmental parameter sensing subsystem, through a distributed sensor network-based global sensing system, can simultaneously capture multi-dimensional environmental characteristics both inside and outside the building. Temperature and humidity gradient data can present temperature stratification and humidity differences between different floors and functional areas; light intensity distribution data can reflect the penetration and obstruction of natural light inside the building; and human activity heatmaps can dynamically track the patterns of crowd gathering and evacuation. The multi-dimensional environmental state vector formed by the fusion of these three data points allows the system to comprehensively grasp the real-time environmental dynamics of the building, avoiding control deviations caused by missing local data.

[0059] The equipment operation status acquisition subsystem comprehensively captures operational data from multiple types of equipment, achieving panoramic coverage of equipment management. The power curves of HVAC units can intuitively present the load change process, the energy efficiency conversion rate curves of lighting equipment groups can reflect the energy utilization level under different operating conditions, and the operational data of renewable energy equipment can provide real-time feedback on clean energy supply capacity. Overlaying the dynamic updates of equipment health indicators, the system can clearly grasp the operational efficiency and health status of each piece of equipment, providing accurate data support for subsequent control.

[0060] The dynamic load forecasting subsystem establishes a spatiotemporal distribution prediction model that, through temporal correlation analysis of historical environmental and equipment data, can predict changes in energy load at different times and in different regions. This prediction is not a simple numerical estimation, but rather a load distribution presentation that combines temporal continuity and spatial differences. For example, it can identify in advance the timing and intensity of load peaks in the morning rush hour office area and the lunch break dining area in commercial buildings, enabling energy allocation to respond to changes in demand in advance.

[0061] The control strategy generation subsystem combines real-time environmental data with an optimized instruction set based on predicted load output to achieve coordinated and refined equipment control. Optimization of start-up and shutdown timing avoids power surges caused by simultaneous start-up and shutdown of multiple devices. Setting power adjustment gradients adapts to the gradual changes in environmental parameters, preventing energy waste and lifespan loss due to frequent device start-ups and shutdowns. Dynamic adjustment of energy allocation weights prioritizes the maximization of renewable energy utilization, such as increasing the proportion of photovoltaic power in lighting supply when photovoltaic output is sufficient, thereby reducing power consumption from the traditional power grid.

[0062] The protocol conversion capability of the execution terminal adaptation subsystem breaks down communication barriers between different devices, enabling optimized commands to be effectively transmitted across brands and types of devices, thus solving the problem of poor device linkage in traditional systems. Simultaneously, dynamic calibration of the actuator's response delay ensures the synchronization of command issuance and device action, avoiding control lags caused by delays. For example, after personnel leave a certain area, lighting and air conditioning can be precisely shut off according to a preset sequence, reducing unnecessary energy consumption. Attached Figure Description

[0063] Figure 1This is a timing diagram of the intelligent control and optimization management system for building equipment based on green and low-carbon buildings as described in this invention.

[0064] Figure 2 Flowchart for generating multidimensional environment state vectors;

[0065] Figure 3 A flowchart for optimizing the generation of control instruction sets. Detailed Implementation

[0066] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0067] Please see Figure 1 This invention provides an intelligent control and optimization management system for building equipment based on green and low-carbon buildings. The system includes: an environmental parameter sensing subsystem, an equipment operating status acquisition subsystem, a dynamic load prediction subsystem, a control strategy generation subsystem, and an execution terminal adaptation subsystem. The environmental parameter sensing subsystem is deployed inside and outside the building via a distributed sensor network. Sensor nodes are arranged in a spatial grid to collect temperature and humidity gradient data, light intensity distribution data, and human activity heatmap data. Temperature and humidity gradient data are measured at multiple points according to building floors and zones. Light intensity distribution data covers the building facade and indoor lighting areas. Human activity heatmap data captures movement trajectories using infrared sensors or video analysis equipment. The equipment operating status acquisition subsystem connects to the controllers of HVAC units, lighting equipment groups, and renewable energy equipment, and reads operating power curves, energy efficiency conversion rate curves, and equipment health indicators in real time. The operating power curve is sampled through an energy monitoring module, the energy efficiency conversion rate curve is calculated based on equipment input and output parameters, and the equipment health indicators integrate multi-source data such as vibration and pressure. The dynamic load forecasting subsystem receives a time-series of historical environmental state vectors and equipment operating status data. The environmental state vectors include timestamps and spatial coordinates, while the equipment operating status data marks load peak events. The subsystem uses machine learning algorithms to build a spatiotemporal distribution prediction model for building energy load, and the model outputs a load distribution map for future time periods. The control strategy generation subsystem integrates the current environmental state vector and the predicted load distribution data. The current environmental state vector reflects real-time parameter deviations, while the predicted load distribution data indicates thermal fluctuations. The subsystem uses optimization algorithms to generate a set of control instructions for equipment start-up and shutdown timing, power regulation gradients, and energy allocation weights, which are stored in matrix form.

[0068] Example 1: See Figure 2The generation process of the multidimensional environmental state vector begins with the acquisition and processing of raw sensor data by the environmental parameter sensing subsystem. The distributed sensor network consists of multiple sensor nodes deployed inside and outside the building. These nodes include temperature and humidity sensors, illuminance sensors, and infrared thermal imaging sensors or video recognition units for sensing human activity. The temperature and humidity sensors are arranged according to the building's spatial topology, which is gridded based on floors, zones, and functional areas defined by the building information model. Each grid cell contains at least one temperature and humidity sensor to collect temperature and humidity gradient data. The collected temperature and humidity gradient data is aggregated to the data preprocessing module of the environmental parameter sensing subsystem via a wireless sensor network protocol. The data preprocessing module filters the raw data to eliminate transient noise interference and marks any data lost due to communication interruptions. The preprocessed temperature and humidity gradient data are subjected to hierarchical clustering analysis based on the building's spatial topology. The hierarchical clustering algorithm uses the physical coordinates of sensor nodes as the clustering basis, grouping spatially adjacent sensor node data into the same cluster. The clustering process uses Euclidean distance to calculate spatial proximity, setting a dynamic distance threshold to ensure that the clustering results reflect the continuity of the building structure. After hierarchical clustering, a spatially continuous temperature and humidity field matrix is ​​formed. The temperature and humidity field matrix is ​​a two-dimensional data structure, where the rows and columns correspond to the coordinates of the building's planar grid. Each element of the matrix stores the temperature and humidity values ​​of the corresponding grid cell. For grid cells containing multiple sensor nodes after clustering, the temperature and humidity values ​​are taken as the arithmetic mean of all sensor readings within the cluster. For grid cells without sensor nodes, the temperature and humidity values ​​are filled using a bilinear interpolation algorithm based on the values ​​of neighboring grid cells, thereby generating a temperature and humidity field matrix that completely covers the building space and has smooth data.

[0069] The processing of light intensity distribution data involves the optical characteristics of the building facade. Light intensity sensors are installed in various lighting areas both outside and inside the building. External light intensity data is convolved with the light transmittance characteristics of the building facade. The light transmittance characteristics of the building facade are a digital model constructed based on factors such as window distribution, glass type, geometric parameters of shading components, and azimuth angles. This model stores the light transmittance coefficients of each point on the building facade in matrix form. The convolution operation uses a convolution kernel customized according to the facade characteristics. The size and weight of the convolution kernel are determined by the window size and light transmittance. The operation simulates the intensity distribution of natural light in the actual indoor space after passing through the facade. The result of the convolution operation generates a corrected indoor light distribution tensor. The indoor light distribution tensor is a three-dimensional data structure whose dimensions include the two-dimensional planar coordinates of the building and the light attenuation levels in the height direction. This tensor accurately reflects the actual light distribution inside the building after the interaction between natural light and the building structure, eliminating the sharp edges and unnatural transitions that may result from directly using sensor data. The heatmap data for personnel activity originates from infrared sensor arrays or privacy-enhanced computer vision analysis systems, which continuously capture the movement and location information of personnel within the building space. Movement trajectory features are extracted by analyzing positional changes across consecutive time frames. The system employs a density clustering algorithm to identify high-frequency activity areas where personnel congregate. These high-frequency activity area coordinate sets consist of a series of planar coordinate points, each associated with a personnel density value. Dwell time sequences are generated by calculating the consecutive occurrence times of the same person near specific coordinate points. The system records a time series for each identified activity area, with each data point representing the average dwell time of a person within a given time window. The high-frequency activity area coordinate sets and dwell time sequences are integrated into a time-stamped personnel distribution layer.

[0070] The temperature and humidity field matrix, the corrected indoor light distribution tensor, and the personnel distribution layer need to be spatiotemporally aligned to be fused into a unified multidimensional environmental state vector. The spatiotemporal alignment operation is based on the GPS timestamp and the spatial coordinate system of the Building Information Model (BIM). The system establishes a unified spatial reference grid, resampling the temperature and humidity field matrix and the indoor light distribution tensor to the same grid precision as the personnel distribution layer. The resampling process uses nearest-neighbor interpolation to ensure that data features are not overly smoothed. Temporal alignment uses a fixed system clock cycle as a reference, synchronizing all data to the same time slice. The aligned data is combined into a structured multidimensional environmental state vector. Each instance of the multidimensional environmental state vector represents a complete snapshot of the building environment at a specific moment. The elements in the vector are arranged sequentially, containing the temperature and humidity values, light intensity values, personnel density values, and timestamp information for each grid cell. The generated multidimensional environmental state vector is stored in a time-series database, with a data format designed to support efficient time range and spatial region queries, providing high-quality, highly consistent input data for the dynamic load forecasting subsystem.

[0071] Example 2: The acquisition of equipment health indicators is performed by the equipment operation status acquisition subsystem. This subsystem establishes communication connections with the local controllers of HVAC units, lighting equipment groups, and renewable energy equipment via hardwired connections or industrial fieldbus. The vibration spectrum characteristics of the HVAC unit compressor are acquired using a piezoelectric accelerometer mounted on the compressor housing. The sensor records vibration signals at a sampling rate of 100,000 times per second. After analog-to-digital conversion, the vibration signals are processed into a spectrum using a fast Fourier transform. The amplitude changes of the fundamental frequency and harmonic components in the spectrum are used to identify rotor imbalance or bearing wear. The refrigerant pressure fluctuation curve is simultaneously monitored by pressure sensors on both the high-pressure and low-pressure sides. The pressure data collected by the pressure sensors is recorded at millisecond intervals, forming a time-series curve. The high-frequency pulsation components on the curve are separated by a low-pass filter, and their fluctuation amplitude is compared and analyzed with a baseline curve under standard operating conditions. The fouling coefficient on the heat exchanger surface is derived by calculating the heat exchange efficiency. The calculation of heat exchange efficiency relies on temperature sensors and flow meter readings installed at the heat exchanger inlet and outlet. The degree of fouling accumulation is quantified by measuring the ratio of actual heat exchange to theoretically clean heat exchange, combined with a model of heat exchanger material and operating time. The voltage harmonic distortion rate of the lighting equipment group is monitored by a smart meter or an embedded power quality analysis module. The monitoring module connects to the total input terminal of the lighting circuit, performs spectral analysis on the collected voltage waveform, and calculates the total harmonic distortion rate and the content of each harmonic. An abnormal increase in the voltage harmonic distortion rate usually indicates aging of the drive power supply or filter circuit. The light source attenuation index is obtained through a photometer or using the lighting equipment's own light feedback sensor. The system periodically records the initial light output value of the lamps under standard nighttime conditions and compares the light output value after a period of operation with the initial value to calculate the luminous flux maintenance curve. The light source attenuation index is derived from the slope of this curve. The change in circuit insulation impedance is measured by periodically injecting low-voltage DC test signals. The insulation monitoring module operates when the equipment is idle, measuring the impedance values ​​between the live wire, neutral wire, and ground wire, and recording their trend over time.

[0072] The cleanliness degradation curve of photovoltaic panels in renewable energy equipment is calculated based on the deviation between real-time power generation and theoretical maximum power. The theoretical maximum power is calculated by combining solar irradiance measured by an irradiance sensor, the nominal efficiency of the photovoltaic panel, and the ambient temperature. The cleanliness degradation coefficient is defined as the ratio of actual power generation to theoretical power. Wind turbine bearing wear is assessed by monitoring the vibration signal characteristics of the wind turbine drivetrain. Vibration acceleration sensors are installed on the gearbox and main shaft bearing housing to analyze the amplitude growth of the component related to the bearing defect frequency in the vibration signal. Inverter efficiency offset is calculated by synchronously measuring the DC-side input power and AC-side output power using a high-precision power sensor, and then comparing the conversion efficiency with the inverter's factory-specified efficiency curve. The characteristic data of these three types of equipment are aggregated into the central processing unit of the equipment operation status acquisition subsystem. All characteristic data includes equipment identifiers and timestamps. The degradation status assessment model uses these feature data as input. It is a multi-layer feedforward neural network trained on historical fault data. The neural network structure includes an input layer, three hidden layers, and an output layer. The number of nodes in the input layer corresponds to the dimension of the feature data, and the output layer nodes output a standardized equipment health index ranging from 0 to 1. A standardized equipment health index closer to 1 indicates a healthier equipment condition, while a index closer to 0 indicates a more severe degradation. After the standardized equipment health index is calculated, it is stored in the equipment status database along with the equipment identifier and timestamp.

[0073] The establishment of the spatiotemporal distribution prediction model is carried out by the dynamic load prediction subsystem. This subsystem accesses a historical database to retrieve historical environmental state vector sequences and equipment operating status data sequences over a certain period. The timestamps of temperature and humidity abrupt change events in the historical environmental state vectors are automatically detected by analyzing the rate of change of the average value of the temperature and humidity field matrix. The system calculates the difference between temperature and humidity values ​​between two consecutive time slices; when the difference exceeds a preset threshold, it is recorded as a timestamp of an abrupt change event. Light intensity transition intervals are identified by monitoring the overall brightness change of the corrected indoor light distribution tensor. The system sets a light intensity change threshold; when the change in light intensity exceeds this threshold within a short period, the time interval is marked as a light intensity transition interval. The corresponding power fluctuation peaks and energy efficiency inflection points in the equipment operating status data need to be matched with the aforementioned environmental events. The matching algorithm uses a sequence alignment technique based on dynamic time warping. Dynamic time warping can overcome the potential time delay between environmental events and equipment responses, accurately associating power fluctuation peaks or energy efficiency inflection points with the timestamps of temperature and humidity abrupt change events or light intensity transition intervals that triggered them. A nonlinear mapping relationship between the rate of change of environmental parameters and the intensity of equipment load response was established using a multivariate nonlinear regression method. The input variables of the mapping model include the rate of change of temperature and humidity, the rate of change of light intensity, and the rate of change of personnel density. The output variable is the change in equipment power. The regression model fits the quantitative relationship between environmental disturbance and load response based on a large amount of historical data.

[0074] A Long Short-Term Memory (LSTM) network is used to construct a spatiotemporal distribution prediction model considering building thermal inertia. LSM is a special type of recurrent neural network structure that includes input, forget, and output gate mechanisms, enabling it to effectively learn long-term dependencies in time series. The network model's input is a sequence of historical environmental state vectors across multiple consecutive time steps. Each environmental state vector contains temperature, humidity, illumination, and occupant data for all grid points within the building, with a sequence length sufficient to cover the period during which building thermal inertia takes effect. The network model's output is a spatiotemporal distribution map of the building's energy load for one or more future time steps, predicting the heating, cooling, and lighting loads for various areas of the building. During the training phase, the LSM uses environmental state vectors from historical data as input and actual measured total power consumption data for subsequent time periods as supervision labels, optimizing network weights through backpropagation. The trained spatiotemporal distribution prediction model can predict the energy load demand of the entire building and its local areas over a future period based on current and historical environmental states. The prediction results are output in matrix form, with the matrix structure consistent with the spatial grid structure of the multidimensional environmental state vectors.

[0075] Example 3: See Figure 3The process of generating optimized control instruction sets is implemented in the control strategy generation subsystem. This subsystem receives the current environmental state vector from the environmental parameter sensing subsystem and the predicted load distribution data from the dynamic load prediction subsystem. Identifying the deviations of anomalous parameters in the current environmental state vector and their spatial impact range is the first step. The deviations are determined by comparing the temperature, humidity, and light intensity values ​​of each grid cell in the current environmental state vector with the upper and lower limits of a preset comfort range. The comfort range is set according to building type and season; for example, the comfortable indoor temperature range in summer is set to 24 to 26 degrees Celsius. For each parameter exceeding the comfort range, its deviation is calculated as the difference between the actual value and the nearest comfort range boundary. The calculation of the spatial impact range is based on the connectivity and airflow model of the building space. An anomalous parameter in one grid cell will affect adjacent areas, with the degree of influence decreasing with increasing distance. The attenuation coefficient is determined by the building's internal partition layout and ventilation conditions. Finally, a spatial distribution map describing the intensity of the anomalous influence is generated.

[0076] The calculation of the safety margin ratio between the predicted load distribution data and the rated capacity of the equipment involves an equipment performance database. The rated capacity of the equipment is obtained from the technical manuals of HVAC units, lighting equipment groups, and renewable energy equipment and stored in the database. The predicted load distribution data provides the cooling, heating, and lighting loads required for each area of ​​the building during a specific future time period. The system sums the area loads to obtain the total load demand. The safety margin ratio is calculated by subtracting the current total load prediction value from the total rated capacity of the equipment, and then dividing by the total rated capacity of the equipment. The formula is expressed as follows:

[0077]

[0078] Where: ξ safe C represents the safety margin ratio. total This indicates the total rated capacity of HVAC units, lighting equipment groups, and renewable energy equipment, in L. predicted This represents the projected total load demand. This ratio is a dimensionless number, ranging from negative infinity to 1. ξ safe A higher value indicates more available capacity and a more secure system operation; safe A negative value indicates that the predicted load has exceeded the total equipment capacity. The equipment health index dynamically adjusts the priority weight coefficients of each device. The initial value of the priority weight coefficient is set according to the importance of the equipment type; for example, HVAC units usually have a higher base weight. The adjustment process monitors the latest equipment health index obtained from the equipment operation status acquisition subsystem in real time. The equipment health index is a standardized value between 0 and 1.

[0079] A multi-objective optimization control instruction set is generated by combining the safety margin ratio and weighting coefficients. This set comprises three core elements: equipment start-up and shutdown sequence, power regulation gradient, and energy allocation weights. The objective function of the optimization problem is defined as minimizing total energy consumption while satisfying building environment comfort constraints. These constraints require that temperature, humidity, and illuminance in all areas be maintained within preset comfort ranges. The equipment start-up and shutdown sequence determines the start-up and shutdown times of each device within a future time window. The power regulation gradient specifies the allowable step size and rate of power adjustment for each device, and the energy allocation weights determine the load distribution ratio among different types of energy. The safety margin ratio ξ... safe It is introduced as a key parameter into the constraint condition when ξ safe When the load falls below a certain threshold, the optimization algorithm prioritizes reducing non-critical loads to ensure system safety. Adjusted priority weight coefficients are introduced into the objective function as weighting factors. Devices with poor health have their weight coefficients reduced, resulting in a decrease in their operational priority during optimization and thus a reduction in their load. The optimization process employs a constrained multi-objective particle swarm optimization algorithm, which outputs a set of optimal device control parameters. These parameters are encoded into a structured set of optimization control instructions.

[0080] The priority weighting coefficient adjustment process is closely coupled with the equipment health index, and the adjustment logic is based on a preset health threshold. When the health index of the HVAC unit is lower than the first threshold, the upper limit of the power regulation gradient of the HVAC unit is reduced. The first threshold is set at 0.7, and the upper limit of the power regulation gradient defines the maximum allowable percentage of a single power adjustment of the HVAC unit. Specifically, if the health index of the HVAC unit is between 0.7 and 0.5, the upper limit of the power regulation gradient is linearly reduced from the normal 10% to 5%; if the health index is lower than 0.5, the upper limit of the power regulation gradient is further limited to 2%. This limitation prevents HVAC units with poor health from being damaged more quickly due to drastic power fluctuations. When the health index of the lighting equipment group is lower than the second threshold, the dimming frequency change rate of the lighting equipment group is limited. The second threshold is set at 0.6, and the dimming frequency change rate refers to the rate of change of brightness of the lighting equipment. For lighting equipment groups with a health index lower than 0.6, the dimming frequency change rate is limited to less than 1% per second to prevent the drive circuit from generating excessive current stress due to frequent rapid dimming. When the health index of renewable energy equipment falls below the third threshold, the energy allocation weight compensation coefficient for the renewable energy equipment is increased. The third threshold is set to 0.8. The energy allocation weight compensation coefficient is a multiplier used to increase the proportion of renewable energy in energy allocation. For example, a renewable energy device with a health index of 0.75 might have its energy allocation weight compensation coefficient set to 1.2. This means that during optimized allocation, the available power of the device will be considered 1.2 times its actual value, encouraging the system to prioritize the use of renewable energy, even if its efficiency decreases, which aligns with the green and low-carbon operation goals. The weight coefficient adjustment module periodically reads the latest equipment health index from the database and updates the priority weight coefficient according to the above rules. The updated coefficient takes effect immediately and is used for the generation of the optimization control instruction set in the next cycle.

[0081] Example 4: The process of converting the optimized control instruction set into control signals compatible with different building equipment protocol stacks is completed by the execution terminal adaptation subsystem, which includes a protocol conversion engine and a timing management module. Parsing the device operation codes and parameter adjustment values ​​in the optimized control instruction set is the first step in the conversion process. The optimized control instruction set exists in the form of structured data objects, which contain the target device identifier, operation instruction type, and specific parameter values. The operation instruction type is encoded as a predefined operation code; for example, operation code "0x01" represents starting the device, and operation code "0x02" represents setting a power percentage. Parameter adjustment values ​​are associated with the operation codes; for example, setting a power percentage operation code must be followed by an integer value between 0 and 100. The parsing engine extracts the operation code and parameter adjustment value required for each control command by traversing the data structure of the optimized control instruction set.

[0082] Matching the communication protocol stack field structure and data encapsulation format of the target device requires querying a device protocol database. This database stores communication protocol information for all controlled devices in the system. Database records include device model, supported protocol types (e.g., ModbusTCP, BACnet / IP, KNX), function codes, register address mapping tables, and data frame encapsulation formats. The protocol conversion engine retrieves the corresponding protocol stack field structure from the device protocol database based on the parsed target device identifier. For example, for an HVAC unit supporting the ModbusTCP protocol, the power setting command might need to be written to holding register address 0x0001, with the data format being a 16-bit unsigned integer. The conversion engine maps common opcodes and parameter adjustments to these device fields and constructs a complete communication data frame according to the data encapsulation format required by the protocol. The data frame includes a frame header, device address, function code, data field, and checksum.

[0083] While converting the opcode into a device-specific instruction set, time-stamp synchronization information is inserted. This time-stamp synchronization information originates from the system's high-precision clock source, which is synchronized with a standard time server via a network time protocol. Before transmission, the constructed device-specific instruction data frame is appended with a precise timestamp, indicating the expected execution time of the instruction. The timing management module maintains an instruction transmission queue, where each instruction carries its own time-stamp synchronization information. The transmission time offset of the control signal is pre-compensated based on the response delay parameter, an estimated value representing the time delay from the transmission of the control signal to the actual start of the device's action. The timing management module uses the response delay parameter to pre-compensate the instruction transmission time, calculated as: Instruction transmission time = Expected instruction execution time - Response delay parameter. This pre-compensation mechanism offsets the inherent delays in network transmission and device processing, ensuring that the device action occurs as close to the expected time as possible. The dynamic calibration of the actuator's response delay parameter is a continuously running background task, relying on precise measurement of the instruction execution cycle. The system records the time difference between the control signal transmission time and the device status feedback reception time. The control signal transmission time is recorded by the execution terminal adaptation subsystem when the command is actually sent to the network, while the device status feedback reception time is recorded when the subsystem receives the device's acknowledgment or status update message. This time difference represents the delay measurement for a complete command-response cycle. Analyzing the delay fluctuation characteristics of different device types under different load rates requires accumulating a large amount of delay measurement data. The system associates and stores each measured time difference with the current device load rate. Load rate information is obtained in real-time from the device operating status acquisition subsystem. Through statistical analysis of historical data, the trend of delay changes with load rate can be observed.

[0084] A piecewise linear regression method is used to establish a dynamic correlation model between equipment response latency and real-time load rate. The model divides the load rate into several intervals and establishes a linear relationship between latency and load rate within each interval. For example, the latency model of an HVAC unit may show that in the load rate range of 0%-30%, the latency remains at a low and stable value; in the load rate range of 30%-70%, the latency increases slowly and linearly with the load rate; and in the load rate range of 70%-100%, the latency may show a faster growth trend. The model parameters are obtained by fitting historical data. The response latency parameter is updated based on the current load rate prediction value, which comes from the dynamic load prediction subsystem. The timing management module periodically (e.g., every minute) queries the load rate prediction value of the main equipment in the next time period and then inputs this prediction value into the corresponding dynamic correlation model between equipment response latency and real-time load rate to calculate an updated response latency parameter prediction value. This new parameter will be used for subsequent control signal transmission time pre-compensation calculations, thereby achieving dynamic adaptation of the latency parameter. Refer to Table 1, which shows a fragment of the protocol mapping relationship.

[0085] Table 1: Mapping Table of Device Operation Codes to Partial Communication Protocols

[0086]

[0087] The execution terminal adaptation subsystem, through the aforementioned process, accurately and promptly converts the abstract instructions generated by the upper-layer optimization strategy into specific control commands that can be recognized and executed by field devices, ensuring the accurate transmission of control intentions and the synchronization of coordinated actions. The collaborative work of the protocol conversion engine and the timing management module effectively solves two key problems: heterogeneous device integration and control timing accuracy.

[0088] Example 5: The energy efficiency assessment subsystem operates independently as a closed-loop feedback link in the management system. The subsystem acquires real-time energy consumption data for HVAC units, lighting equipment groups, and renewable energy equipment from the equipment operation status acquisition subsystem via a data interface. The energy consumption data is measured in kilowatt-hours (kWh) and includes a timestamp and equipment identifier. Real-time calculation of the ratio of energy intensity per unit area to carbon emission intensity for each equipment group requires the integration of building space information, extracted from the building information model and including area data for each functional zone. The calculation of energy intensity per unit area involves dividing the total energy consumption of a single equipment group over a specified time period by the building area of ​​the area it serves. For example, the total power consumption of the lighting equipment group in an office area is divided by the total area of ​​the office area, resulting in an intensity value in kilowatt-hours per square meter. The calculation of the carbon emission intensity ratio requires the introduction of the carbon emission factor of energy. The carbon emission factor is determined based on the energy structure of the local power grid or the data of the gas supplier. The energy consumption of the equipment is multiplied by the corresponding carbon emission factor to obtain the absolute carbon emission amount, which is then divided by the building area of ​​the service area to obtain the carbon emission intensity per unit area. Finally, this intensity value is divided by a preset benchmark intensity value to obtain the carbon emission intensity ratio.

[0089] The calculation results are compared item by item with the thresholds of green building certification standards. These thresholds are derived from authoritative green building assessment systems, such as LEED or the China Green Building Evaluation Standard. These thresholds are stored in the subsystem database as key performance indicators (KPIs), including the upper limit of annual electricity consumption per unit area and the upper limit of annual carbon emissions per unit area. The comparison process is automated. The system compares the calculated ratio of energy intensity and carbon emission intensity per unit area with the target thresholds of the corresponding indicators in the database, recording the magnitude and direction of the deviation. An energy efficiency deviation correction coefficient is generated and fed back to the control strategy generation subsystem. The energy efficiency deviation correction coefficient is a numerical correction factor, and its generation logic is based on the comparison results. When the calculated value is better than the threshold, the correction coefficient may be 1 or less, indicating that no correction is needed or control can be appropriately relaxed. When the calculated value is worse than the threshold, the correction coefficient is greater than 1, indicating that energy-saving control needs to be strengthened. This coefficient is encapsulated into a standardized data packet and sent to the control strategy generation subsystem in real time via the system's internal communication bus, serving as an important weight parameter in its optimization objective function. The generation process of the energy efficiency deviation correction coefficient includes detailed diagnostic analysis. The system identifies equipment types and their spatial distribution that exceed benchmark values ​​for energy consumption per unit area. These benchmark values ​​are typically set as industry standards or the building's historical best operating data. The system periodically calculates the energy consumption per unit area for each equipment group and compares it to the set benchmark values. For equipment types exceeding the benchmark values, the system further analyzes the spatial distribution characteristics of their high energy consumption. For example, by overlaying geographic information system layers, it identifies abnormally high energy consumption of HVAC units in the building's north area. The system also analyzes equipment operation patterns in areas with abnormal carbon emission intensity ratios. These areas are defined as those with carbon emission intensity per unit area significantly higher than the building average. The system retrieves the operating logs of all relevant equipment in these areas during the abnormal time period and analyzes their operation patterns. For example, it finds that when office lighting equipment is fully operational and HVAC unit settings are too low, the carbon emission intensity ratio in that area rises sharply. This pattern analysis helps identify the interconnected relationships between equipment leading to low energy efficiency.

[0090] The system excludes non-fault-related energy efficiency deviations based on equipment health indicators, which are derived from the equipment operating status acquisition subsystem. This is a crucial filtering step designed to avoid incorrectly attributing energy inefficiencies caused by equipment performance degradation to the control strategy. For example, if an air handling unit experiences increased energy consumption per unit area, but its equipment health indicators show severe fan bearing wear and decreased efficiency, this energy increase is considered non-fault-related. The generated energy efficiency deviation correction coefficient will be specially marked or use different calculation logic, potentially emphasizing recommended maintenance rather than aggressive control parameter adjustments. Generating energy efficiency deviation correction coefficients for specific equipment combinations is the final output step. These coefficients are no longer simple global scalars but are tailored to the identified low-energy-efficiency equipment combination patterns. For example, for the pattern of "all office lighting on + air conditioning set to low temperature," the system generates a correction coefficient for this specific combination. This coefficient simultaneously affects the optimization weights of the lighting and air conditioning control modules in the control strategy generation subsystem, forcing the optimization algorithm to prioritize breaking this high-carbon combination pattern during calculation, for example, by automatically dimming the lighting in unoccupied areas when the air conditioning load is high. The energy efficiency assessment subsystem, through this refined, diagnostic-based coefficient generation mechanism, enables more targeted adjustments to control strategies, thereby more effectively driving the improvement of the overall energy efficiency level of the system and meeting the long-term operational goals of green and low-carbon buildings.

[0091] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0092] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A building equipment intelligent control and optimization management system based on green and low-carbon buildings, characterized in that, include: The environmental parameter sensing subsystem collects temperature and humidity gradient data, light intensity distribution data, and human activity heat map data inside and outside the building through a distributed sensor network to form a multi-dimensional environmental state vector. The equipment operation status acquisition subsystem acquires real-time operating power curves, energy efficiency conversion rate curves, and equipment health indicators for HVAC units, lighting equipment groups, and renewable energy equipment. The dynamic load forecasting subsystem establishes a spatiotemporal distribution forecasting model for building energy load based on the temporal correlation between historical environmental state vectors and equipment operating status data. The control strategy generation subsystem combines the current environmental state vector with the load distribution data output by the prediction model to generate an optimized control instruction set that includes equipment start-up and shutdown timing, power regulation gradient, and energy allocation weights. The execution terminal adaptation subsystem converts the optimized control instruction set into control signals compatible with different building equipment protocol stacks and dynamically calibrates the response delay parameters of the actuators.

2. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 1, characterized in that, The process of generating the multidimensional environment state vector is as follows: The temperature and humidity gradient data are hierarchically clustered according to the building spatial topology to form a temperature and humidity field matrix with spatial continuity; The light intensity distribution data is convolved with the light transmittance characteristics of the building facade to generate the corrected indoor light distribution tensor. Based on the movement trajectory characteristics of personnel activity heatmap data, extract the coordinate set of high-frequency activity areas and the stay duration sequence; After aligning the temperature and humidity field matrix, the modified illumination distribution tensor, and the personnel coordinate set in time and space, a multidimensional environmental state vector is constructed.

3. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 1, characterized in that, The process for obtaining the equipment health index is as follows: The vibration spectrum characteristics of the HVAC unit compressor, the refrigerant pressure fluctuation curve, and the fouling coefficient of the heat exchanger surface were collected. Monitor the voltage harmonic distortion rate, light source attenuation index, and circuit insulation impedance changes of lighting equipment groups; Analyze the cleanliness degradation curve of photovoltaic panels, the wear of wind turbine bearings, and the efficiency offset of inverters in renewable energy equipment. By combining the characteristic data of the three types of equipment, a standardized equipment health index is output through a degradation status assessment model.

4. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 1, characterized in that, The process of establishing the spatiotemporal distribution prediction model is as follows: Extract the timestamps of temperature and humidity abrupt events and the light intensity transition intervals from the historical environmental state vector; Match the peak power fluctuations and energy efficiency inflection points in the equipment's operating status data; Establish a nonlinear mapping relationship between the rate of change of environmental parameters and the intensity of equipment load response; A spatiotemporal distribution prediction model considering building thermal inertia is constructed using a long short-term memory network.

5. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 1, characterized in that, The process of generating the optimized control instruction set is as follows: Identify the deviation of abnormal parameters in the current environmental state vector and their spatial influence range; Calculate the safety margin ratio between the predicted load distribution data and the rated capacity of the equipment; The priority weight coefficients of each device are dynamically adjusted based on the device health index. An optimized control instruction set containing multi-objective constraints is generated by combining the safety margin ratio and weighting coefficients.

6. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 5, characterized in that, The adjustment process for the priority weight coefficient is as follows: When the health index of the HVAC unit is lower than the first threshold, reduce its power regulation gradient upper limit. When the health index of a group of lighting devices falls below the second threshold, the rate of change of its dimming frequency is limited. When the health index of renewable energy equipment falls below the third threshold, its energy allocation weight compensation coefficient is increased.

7. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 1, characterized in that, The process of converting the optimized control instruction set into control signals compatible with different building equipment protocol stacks is as follows: Analyze and optimize the device operation codes and parameter adjustment values ​​in the control instruction set; Match the communication protocol stack field structure and data encapsulation format of the target device; While converting the opcode into a device-specific instruction set, time stamp synchronization information is inserted; The transmission time offset of the control signal is pre-compensated based on the response delay parameter.

8. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 7, characterized in that, The response delay parameter process of the dynamic calibration actuator is as follows: Record the time difference between the control signal transmission time and the device status feedback reception time; Analyze the latency fluctuation characteristics of different equipment types under different load rates; Establish a dynamic correlation model between device response latency and real-time load rate; Update the response latency parameters based on the current load rate forecast.

9. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 1, characterized in that, Also includes: The energy efficiency assessment subsystem calculates the ratio of energy consumption intensity to carbon emission intensity per unit area for each equipment group in real time. The calculation results were compared item by item with the thresholds of the green building certification standard. The energy efficiency deviation correction coefficient is generated and fed back to the control strategy generation subsystem.

10. The intelligent control and optimization management system for building equipment based on green and low-carbon buildings according to claim 9, characterized in that, The process for generating the energy efficiency deviation correction coefficient is as follows: Identify the types of equipment whose energy consumption per unit area exceeds the benchmark value and their spatial distribution; Analyze the equipment operation combination patterns in areas with abnormal carbon emission intensity ratios; Eliminate non-fault performance deviations based on equipment health indicators; Generate energy efficiency deviation correction coefficients for specific equipment combinations.

Images