A smart home digital management system based on the Internet of Things

Abstract

The application discloses a smart home digital management system based on the Internet of Things, relates to the field of the Internet of Things and building informatization, and comprises a sensing layer, a network layer, a platform layer and an application layer. The edge gateway of the network layer runs closed-loop control according to an environment, a process, a progress state vector in a construction stage. When a completion delivery trigger event is detected, initial model parameters of an operation and maintenance stage are calculated according to model parameters at the end of the construction stage, preset default parameters of the system and an adaptive weight coefficient, the closed-loop control of the operation and maintenance stage is switched to run according to a room environment state vector, and the hardware of the edge gateway and the modeling framework remain unchanged. The edge gateway is further provided with semantic mapping middleware, and unified mapping and conflict resolution of heterogeneous protocol devices are realized based on a knowledge graph. The application realizes whole-cycle cooperation of construction and operation and maintenance and cross-cycle reuse of hardware.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, specifically to a smart home decoration digital management system based on IoT, belonging to the interdisciplinary field of IoT, smart home control and building information technology. Background Technology

[0002] The Internet of Things (IoT) technology enables environmental status perception and data transmission by deploying sensors and wireless communication modules. Edge computing gateways are deployed on the field side and have protocol conversion and local data processing capabilities, which can reduce cloud communication latency and are widely used in smart home and industrial field monitoring.

[0003] Building Information Modeling (BIM) technology uses three-dimensional digital models to carry building spatial attributes, component parameters, and construction progress information, providing data support for the information management of the entire building life cycle. Currently, the most widely used digital management solution for home decoration consists of three independent parts: a quotation and design system, a manual management method at the construction site, and a smart home system after occupancy. At the construction site, human experience is used to schedule personnel, material arrival, and environmental conditions. On-site testing data is only for manual reference and is not linked with equipment.

[0004] After moving in, the smart home system triggers device start-up and shutdown based on fixed threshold rules of a single environmental parameter, lacks the ability to perform multi-variable collaborative optimization, and does not receive physical characteristic data of the house generated during the construction phase;

[0005] The existing technology in this field generally believes that IoT devices in the construction phase and smart home devices in the operation and maintenance phase should be deployed independently due to differences in business cycle, monitoring objectives and modeling methods. Devices in the construction phase are usually dismantled and abandoned after completion and acceptance, forming a technical convention that "devices in the construction phase and devices in the operation and maintenance phase must be configured independently". The above convention leads to the following defects: the strong coupling relationship between multiple variables in the construction phase cannot be automatically linked and adjusted, resulting in potential quality problems and project delays.

[0006] The data on concealed works collected during the construction phase were not transmitted to the operation and maintenance phase, resulting in a crude operation and maintenance control strategy and limited energy efficiency.

[0007] The communication protocols of devices from different manufacturers are incompatible with each other. Existing gateways can only complete physical layer protocol bridging and lack the ability to perform unified semantic layer mapping and control conflict resolution, which makes it impossible for the system to make automatic decisions when there are conflicts in the collaborative goals of devices. Summary of the Invention

[0008] The technical problem this invention aims to solve is that existing digital home decoration systems separate construction management from smart home control after occupancy into unrelated systems. During the construction phase, there is a lack of dynamic collaborative control capabilities for multiple variables such as personnel, materials, and environment. After completion, the IoT hardware and the hidden engineering data it collects cannot be reused. Furthermore, the semantics between heterogeneous communication protocol devices are not interoperable, and control conflicts cannot be automatically resolved, resulting in frequent hidden quality problems in the project, crude control strategies during the operation and maintenance phase, and low multi-device collaboration capabilities.

[0009] To address the aforementioned technical issues, this invention provides a smart home decoration digital management system based on the Internet of Things (IoT). This system comprises a perception layer, a network layer, a platform layer, and an application layer. The perception layer includes environmental sensors deployed at the construction site, BLE beacons worn by workers, and RFID tags affixed to building materials during the construction phase. During the operation and maintenance phase, it includes smart home execution devices and supporting sensors. The network layer includes an edge gateway, which integrates a multi-protocol RF transceiver module, an edge computing unit, a local storage module, and an MQTT communication module. The platform layer includes a cloud-based microservice module, a physical model library, and a graph database. The graph database is used to integrate and store building information model spatial attribute data, construction progress data, and IoT time-series data. The application layer includes a construction scheduling dashboard, a supplier logistics collaboration interface, and a smart home control terminal.

[0010] The core of this invention lies in the fact that the edge computing unit performs closed-loop control of the construction phase based on state vectors including regional temperature and humidity, formaldehyde concentration and process completion, as well as control vectors including dehumidification equipment start-up and shutdown, heating equipment start-up and shutdown, and personnel scheduling early warning signals.

[0011] During the operation and maintenance phase, closed-loop control is performed based on state vectors including room temperature and humidity and air quality indicators, as well as control vectors including start-stop and speed commands for air conditioning, fresh air equipment and underfloor heating equipment.

[0012] The edge computing unit continuously monitors the completion and acceptance status of the work process. When it detects a completion and delivery trigger event where the completion of the work process reaches a preset threshold and the acceptance is qualified, it calculates the initial model parameters for the operation and maintenance phase by weighting the model parameters accumulated at the end of the construction phase, the default model parameters preset by the system, and the adaptive weight coefficients calculated based on the completeness of construction data and the acceptance rate. Then, it switches the operation mode from closed-loop control in the construction phase to closed-loop control in the operation and maintenance phase. During the switch, the hardware and state space modeling framework of the edge gateway remain unchanged. In addition, the edge gateway also deploys semantic mapping middleware to realize unified access mapping and control conflict resolution of heterogeneous protocol devices based on a unified state dictionary and knowledge graph.

[0013] The beneficial effects of this invention are:

[0014] 1. By sharing a unified state-space modeling framework between the construction and operation and maintenance phases and achieving adaptive and smooth inheritance of model parameters, the environmental, technological, and schedule variables during the construction phase can be dynamically and collaboratively adjusted, reducing the probability of potential engineering quality problems and project delays.

[0015] 2. Existing technologies are limited by the convention that "construction and maintenance equipment must be deployed independently," requiring the re-purchase and deployment of IoT equipment during the maintenance phase after completion. This invention breaks through the above convention by enabling the edge gateway to complete the role transformation before and after completion and delivery without disassembly or replacement, simply by switching the software mode of the edge computing unit. This achieves the reuse of IoT hardware resources throughout the entire life cycle and reduces the cost of repeated deployment.

[0016] 3. By transforming concealed engineering data into prior parameters of the state transition prediction function during the operation and maintenance phase, the precision and energy efficiency of the environmental control strategy after occupancy are improved.

[0017] 4. By using a knowledge graph-based semantic adaptive mapping middleware to automatically learn semantic nodes for newly connected devices, and using the weighted sum of the deviations between total energy consumption and environmental quality indicators as the optimization objective to solve the joint control vector of conflicting devices, unified semantic access and multi-objective collaborative decision-making for heterogeneous protocol devices are realized, eliminating information silos and improving the collaborative efficiency of multiple devices. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the overall system architecture of the present invention.

[0019] Figure 2 This is a flowchart of the cross-period multivariable collaborative control method of the present invention;

[0020] Figure 3 This is a flowchart of the heterogeneous protocol semantic mapping and conflict resolution process of the present invention. Detailed Implementation

[0021] The following is in conjunction with the appendix Figure 1-3 The specific embodiments of the present invention will be further described below. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0022] I. System Overall Architecture

[0023] The smart home decoration digital management system based on the Internet of Things provided in this embodiment is divided into a perception layer, a network layer, a platform layer, and an application layer in terms of architecture, and uses the edge gateway as the core link to run through the two business cycles of construction and operation and maintenance.

[0024] The perception layer, during the construction phase, includes multi-functional environmental sensors deployed on the construction site (for collecting temperature, humidity, TVOC, and PM2.5 data), BLE beacons worn by workers, and RFID tags affixed to key building materials. During the operation and maintenance phase, it includes smart home actuators (air conditioners, fresh air systems, and underfloor heating systems) and supporting sensors. The network layer is carried by an edge gateway, which integrates multi-protocol RF transceiver modules, supports device access via protocols such as BLE, Zigbee 3.0, and Wi-Fi 6, and maintains a long-term connection with the cloud platform through an MQTT communication module. The platform layer adopts a microservice architecture, defines a standardized object model, and integrates and stores the spatial attribute data (IFC format) of the building information model, construction progress data (Gantt chart data), and IoT time-series data through a graph database. The application layer provides construction scheduling dashboards for decoration companies, logistics collaboration interfaces for suppliers, and smart home control terminals for homeowners.

[0025] II. Edge Gateway Hardware Structure

[0026] The edge gateway, as the core hardware carrier of this invention's cross-cycle collaboration, internally includes a multi-protocol RF transceiver module, a sensor and actuator access interface, an edge computing unit, a local storage module, an MQTT communication module, and a power supply module. The multi-protocol RF transceiver module is equipped with three sub-units: BLE, Zigbee 3.0, and Wi-Fi 6, which are connected to the edge computing unit via an internal bus. The sensor and actuator access interface connects to various field sensors and actuators. The edge computing unit internally runs two functional sub-modules: a multivariable collaborative control algorithm module and a semantic mapping middleware module. The local storage module is bidirectionally connected to the edge computing unit and is used to cache state space model parameters and local copies of the knowledge graph to ensure that local closed-loop control can still be maintained when the cloud connection is interrupted. The MQTT communication module connects the edge computing unit and the cloud platform. The power supply module provides unified power to all the above modules.

[0027] III. Cross-Period Multivariable Dynamic Cooperative Control Method

[0028] One of the core methods of this invention is a cross-cycle multivariable collaborative control method based on the concept of model predictive control. This method uses the same state-space modeling framework in both the construction and operation and maintenance phases, and only switches and inherits the content and parameters of the predictive model according to the phase.

[0029] Define the system in The state vector at time t is:

[0030] ;

[0031] in, express The column vector formed by the controlled variables of the system at each time step. Indicates the first Each controlled variable in The values ​​collected at specific times include temperature and humidity, formaldehyde concentration, and completion status of each area during the construction phase; and temperature, humidity, and air quality indicators of each room during the operation and maintenance phase. The total number of controlled variables, indicated by the superscript. This represents the transpose of a vector.

[0032] Define the system in The control vector at time t is:

[0033] ;

[0034] in, express A column vector consisting of control commands constantly sent to each execution device. Indicates the first Each execution device in The control parameters at any given time include, during the construction phase, the start / stop of dehumidification equipment and heating equipment, as well as personnel dispatch warning signals; during the operation and maintenance phase, they include the start / stop and speed control commands for air conditioning, fresh air systems, and underfloor heating systems. To control the total number of variables.

[0035] The system prediction model is as follows:

[0036] ;

[0037] in, The state transition prediction function is constructed using empirical formulas for material curing during the construction phase (e.g., the relationship between paint drying rate and temperature and humidity), and during the operation and maintenance phase, it is constructed using a building thermodynamic model with prior parameters of concealed works (insulation coefficient, pipeline coordinates). express The external disturbance vector at any given time includes weather forecast data and the original construction schedule data.

[0038] The cost function for system optimization is:

[0039] ;

[0040] in, This represents the total cost within the prediction time domain. Indicates the length of the prediction time domain. Indicates the first The reference trajectory at any given moment is derived from the process library (the range of maintenance environment requirements during the construction phase) or the comfort library (the range of human comfort during the operation and maintenance phase). This represents the state deviation weight matrix, used to measure the cost of deviating from the reference trajectory. This represents the control energy consumption weight matrix, used to measure the energy cost of performing control actions.

[0041] The edge computing unit performs the following processing at each sampling time: It collects the current state vector. And read the reference trajectory from the process library or comfort library. At the same time, acquire the future external perturbation prediction vector sequence of the step The corresponding state transition prediction function is called based on the current stage of the system. Build a predictive model, and based on this predictive model, make predictions about the future. The state trajectories at each time step are recursively predicted to obtain the predicted state sequence, using the cost function. The optimization objective is to reduce the value of the target value. Under the condition of satisfying the physical constraints of the execution equipment (such as start / stop status and gear range), a quadratic programming method is used to solve for the future value. The control sequence of the step sends the optimal control quantity obtained at the current moment to the field execution device to drive it to perform the corresponding action. At the next sampling moment, the actual state of the system is re-acquired and used as the new initial state to repeat the above process, forming a rolling closed-loop control.

[0042] The parameter inheritance function during stage transition is:

[0043] ;

[0044] in, This represents the initial set of model parameters during the operation and maintenance phase, including the weight matrix and prior parameters for hidden engineering. This represents the set of model parameters representing the sedimentation at the end of the construction phase. This represents the set of pre-set general operation and maintenance model parameters of the system. This represents the adaptive weighting coefficient, and its value range is... It is dynamically calculated from indicators such as the completeness of construction data and the acceptance rate. The larger the value, the greater the contribution of the data accumulated during the construction phase to the initial parameters during the operation and maintenance phase.

[0045] The edge computing unit continuously monitors the completion and acceptance status of processes. When it detects a completion and delivery trigger event indicating that the process completion has reached a preset threshold and the acceptance is qualified, it calculates the initial model parameters for the operation and maintenance phase based on the aforementioned parameter inheritance function. Furthermore, the system operation mode is switched from a closed loop during the construction phase to a closed loop during the operation and maintenance phase, while retaining the edge gateway hardware and the underlying state space model framework unchanged, thus achieving a smooth transfer of control. Since the data of concealed works (coordinates of water and electricity pipelines, actual construction parameters of wall insulation materials) are collected and stored in the graph database during the construction phase, the above data can be directly loaded as prior parameters when constructing the state transition prediction function during the operation and maintenance phase, thereby improving the refinement of the environmental control strategy during the operation and maintenance phase.

[0046] IV. Heterogeneous Protocol Semantic Adaptive Mapping and Conflict Resolution Middleware

[0047] Another core method of this invention is a semantic mapping middleware deployed within an edge gateway, used to resolve semantic interoperability and control conflicts between heterogeneous communication protocol devices. This middleware maintains a unified state dictionary. ,in The first one in the unified state dictionary represents the first one. A standardized semantic field, To determine the total number of fields, a knowledge graph is maintained simultaneously. ,in It is a collection of nodes, and the node types include device nodes, spatial nodes, and semantic concept nodes. Let be the set of edges, representing the hierarchical, positional, and semantic relationships between nodes.

[0048] The semantic mapping middleware continuously listens to the raw communication packets reported by devices using various heterogeneous protocols (Zigbee, Bluetooth Mesh, Wi-Fi), extracts feature information such as device identification fields, data type fields, and numerical fields from the packets, and then compares the extracted feature information with the knowledge graph. The system performs similarity matching on existing device semantic nodes, based on device model database records and historical mapping records. If a match is successful, the device data is directly mapped to a unified state dictionary. The corresponding standardized fields If a match fails, it is determined to be a newly connected device. Then, based on the device's spatial topology location (room or area) and the range and cycle characteristics of its reported data, combined with preset rules, the process is inferred within the knowledge graph. Add corresponding device semantic nodes and associated edges, and update the unified state dictionary synchronously. Field definitions.

[0049] Middleware continuously monitors the unified state dictionary For the multi-device control requests corresponding to each field, when two or more devices in the same space make conflicting control requests for related environmental objectives (e.g., the fresh air system requests to start ventilation and the air conditioning system requests to maintain airtightness and insulation simultaneously), a conflict resolution process is triggered. If no conflict is detected, the control requests of each device are issued normally according to the original instructions. For the set of devices that have detected conflicts, a joint optimization function based on the deviation of overall energy consumption and environmental quality indicators is constructed, as follows:

[0050] ;

[0051] in, This represents the joint control vector composed of the control variables of the conflicting devices. This represents the total energy consumption function corresponding to the joint control vector. Indicated in the joint control vector Under the action of the first The actual values ​​of environmental quality indicators (such as air quality, temperature and humidity comfort, or construction environment quality indicators). Indicates the first The target values ​​for each environmental quality indicator, This represents the total number of environmental quality indicators. This represents the trade-off coefficient between energy consumption costs and environmental quality deviations. By solving the above optimization function, we obtain the joint control vector that reduces the value of the objective function. The middleware will The process involves breaking down the conflict into specific execution instructions for each conflicting device and issuing them. Simultaneously, the device combinations, spatial locations, and final solutions for this conflict scenario are recorded in a knowledge graph. In this process, historical cases are used for matching and reference in subsequent similar conflict scenarios, forming a closed loop of continuous optimization. Example Description

[0052] Taking a residential home renovation project as an example: During the construction phase, the edge gateway collects temperature, humidity, formaldehyde concentration, and process completion status data for each area of ​​the construction site to form a state vector. Combined with weather forecast data and the original construction schedule, this forms an external disturbance vector. Based on the relationship between paint drying rate and temperature / humidity, a prediction model is constructed. Under the condition of satisfying the start / stop constraints of dehumidification and heating equipment, the control sequence is solved, and the current control quantity is issued, achieving dynamic linkage between the wall curing environment and construction processes. When the edge gateway detects that the completion status of all processes has reached a preset threshold and the acceptance is qualified, it uses model parameters accumulated at the end of the construction phase, system preset default parameters, and adaptive weight coefficients calculated based on the completeness of construction data and the acceptance rate. The initial model parameters for the operation and maintenance phase are calculated, and the operation mode is switched to closed-loop control for the operation and maintenance phase. At this time, the edge gateway continues to operate with the original hardware, collects temperature, humidity and air quality indicators of each room to form a new state vector, and constructs a building thermodynamic prediction model by combining the coordinates of water and electricity pipelines and the wall insulation coefficient collected during the construction phase. The start-stop and speed commands of air conditioning, fresh air equipment and underfloor heating equipment are solved and issued. During this process, the semantic mapping middleware continuously completes semantic node learning and unified state dictionary mapping for newly connected smart home devices. When a conflict is detected between the control objectives of fresh air equipment and air conditioning equipment, the joint control vector after resolving the conflict is obtained by solving the joint optimization function and then issued for execution.

[0053] Those skilled in the art should understand that the above embodiments are merely illustrative examples of the present invention and are not intended to limit the scope of protection of the present invention. Equivalent substitutions or improvements made without departing from the essence of the technical solution of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A smart home decoration digital management system based on the Internet of Things, comprising a perception layer, a network layer, a platform layer, and an application layer; The sensing layer includes environmental sensors deployed at the construction site, BLE beacons worn by workers, and RFID tags affixed to building materials during the construction phase, and smart home execution devices and supporting sensors during the operation and maintenance phase. The network layer includes an edge gateway, which integrates a multi-protocol radio transceiver module, an edge computing unit, a local storage module, and an MQTT communication module. The multi-protocol radio transceiver module is used to access devices with different communication protocols in the perception layer, and the MQTT communication module is used to establish a communication connection with the platform layer. The platform layer includes a cloud microservice module, a physical model library, and a graph database. The graph database is used to integrate and store building information model spatial attribute data, construction progress data, and Internet of Things time-series data. The application layer includes a construction scheduling dashboard, a supplier logistics collaboration interface, and a smart home control terminal. Its features ； The edge computing unit is configured to run closed-loop control during the construction phase based on state vectors including regional temperature and humidity, formaldehyde concentration, and process completion, as well as control vectors including dehumidification equipment start / stop, heating equipment start / stop, and personnel dispatch early warning signals. During the operation and maintenance phase, it runs closed-loop control based on state vectors including room temperature and humidity and air quality indicators, as well as control vectors including air conditioning, fresh air equipment, and floor heating equipment start / stop and gear commands. The edge computing unit is also configured to continuously monitor the completion and acceptance status of the work process. When a completion and delivery trigger event is detected, indicating that the completion of the work process has reached the completion threshold set according to the construction acceptance standards and the acceptance is qualified, the initial model parameters for the operation and maintenance phase are obtained through weighted calculation based on the model parameters accumulated at the end of the construction phase, the default model parameters preset by the system, and the adaptive weight coefficients calculated based on the completeness of construction data and the acceptance pass rate. The operating mode is then switched from the closed-loop control of the construction phase to the closed-loop control of the operation and maintenance phase. During the switching process, the hardware and state space modeling framework of the edge gateway remain unchanged, and the switching is completed only by adjusting the software configuration of the operating parameters and state vector dimensions within the edge computing unit, without the need for hardware disassembly, replacement, or redeployment of the edge gateway.

2. The system according to claim 1, characterized in that, The edge computing unit is configured to perform the following processing: Obtain the current state vector and the reference trajectory of the corresponding stage, and obtain the external disturbance prediction sequence for a set number of future time steps, wherein the set number of time steps is determined based on the available duration of the external disturbance prediction data; The corresponding state transition prediction function is invoked according to the current stage. The state transition prediction function is constructed based on the material solidification empirical formula in the construction stage and based on the building thermodynamic model and loaded with the prior parameters of the hidden works in the operation and maintenance stage. Based on the state transition prediction function, the state trajectory of a preset number of future time steps is recursively predicted to obtain the predicted state sequence. The optimization objective is to reduce the weighted sum of the deviation between the predicted state sequence and the reference trajectory, as well as the energy consumption cost corresponding to the control vector, to solve for the control sequence for a preset number of future time steps. The control quantity corresponding to the current moment obtained from the solution is sent to the execution device.

3. The system according to claim 2, characterized in that, The edge computing unit is further configured to: reacquire the actual system state vector at the next sampling time as a new initial state, and repeat the processing described in claim 2 to form rolling closed-loop control.

4. The system according to claim 1, characterized in that, The edge gateway is also deployed with a semantic mapping middleware, which maintains a unified state dictionary and a knowledge graph. The knowledge graph includes device nodes, spatial nodes, and semantic concept nodes, as well as edges representing the subordinate relationships, positional relationships, and semantic associations between nodes. The semantic mapping middleware is configured to extract the device identifier field, data type field, and numerical field from the reported messages of heterogeneous protocol devices, match the extracted feature information with the device semantic nodes in the knowledge graph, and map the corresponding device data to the standardized fields in the unified state dictionary when a match is successful.

5. The system according to claim 4, characterized in that, The semantic mapping middleware is also configured to, when a match fails, add corresponding device semantic nodes and associated edges in the knowledge graph based on the spatial topology location of the newly accessed device, the value range of the reported data, and the change cycle characteristics, and update the field definitions of the unified state dictionary.

6. The system according to claim 4, characterized in that, The semantic mapping middleware is also configured to continuously monitor multiple device control requests corresponding to each field in the unified state dictionary. When it detects that two or more devices in the same space have made conflicting control requests for related environmental targets, it constructs a joint control vector optimization problem with the weighted sum of the deviations of total energy consumption and environmental quality indicators as the optimization objective function. By solving the optimization objective function, it obtains a joint control vector that reduces the weighted sum. The joint control vector is decomposed into execution instructions corresponding to each conflicting device and issued. The device combination, spatial location and execution instructions of this conflict scenario are recorded in the knowledge graph.

7. The system according to claim 2, characterized in that, When the edge computing unit constructs the state transition prediction function during the operation and maintenance phase, it loads the concealed works data collected and stored in the graph database during the construction phase as prior parameters. The concealed works data includes the coordinates of water and electricity pipelines and the construction parameters of wall insulation materials.

8. The system according to claim 1, characterized in that, The edge gateway also includes a sensor and execution device access interface and a power module. The sensor and execution device access interface connects to field sensors and execution devices. The local storage module is used to cache state space model parameters and local copies of the knowledge graph. The power module supplies power to the multi-protocol radio frequency transceiver module, the edge computing unit, the local storage module, and the MQTT communication module.

9. A smart home decoration cross-cycle multi-variable collaborative control method based on the Internet of Things, applied to the system described in claim 1, the method comprising: During the construction phase, state vectors including regional temperature and humidity, formaldehyde concentration, and process completion are collected. Based on the empirical formula for material curing, a state transition prediction function is constructed to predict the state trajectory for the next preset time step. With the goal of weighting and reducing the deviation between the predicted state trajectory and the reference trajectory in the process library and the cost of control energy consumption, the control vectors including the start and stop of dehumidification equipment, the start and stop of heating equipment, and personnel scheduling early warning signals are obtained. The control quantity corresponding to the current moment is then sent to the execution equipment. Continuously monitor the completion and acceptance status of the process. When a completion and delivery trigger event is detected, the initial model parameters for the operation and maintenance phase are obtained by weighted calculation based on the model parameters accumulated at the end of the construction phase, the default model parameters preset by the system, and the adaptive weight coefficients calculated based on the completeness of construction data and the acceptance pass rate. During the operation and maintenance phase, the initial model parameters of the operation and maintenance phase are loaded, and state vectors including room temperature, humidity and air quality indicators are collected. Based on the building thermodynamic model and the state transition prediction function constructed by loading the prior parameters of the concealed works, the state trajectory of the future preset time steps is predicted. With the goal of weighting and reducing the deviation between the predicted state trajectory and the reference trajectory in the comfort library and the control energy consumption cost, the control vector including the start-stop and gear commands of the air conditioning, fresh air equipment and underfloor heating equipment is solved, and the control quantity corresponding to the current moment is sent to the execution equipment.

Images