Health monitoring method and system for intelligent building

Abstract

The invention discloses a health monitoring method and system for an intelligent building, and belongs to the field of building construction health monitoring, and the method comprises the steps: obtaining internal and external multi-dimensional data of the intelligent building in real time through a plurality of sensors, the multi-dimensional data comprising environment data, structure data of the intelligent building, equipment state data and personnel activity data; preprocessing the multi-dimensional data to obtain preprocessed multi-dimensional data; according to the preprocessed multi-dimensional data, a health condition analysis result of the intelligent building is output by utilizing a pre-trained health monitoring model, and the health condition analysis result comprises the suitability degree of environmental conditions, the stability degree of a building structure, the equipment health degree and the risk degree of personnel activities; when the health condition analysis result indicates that the intelligent building is abnormal in health, an early warning signal is sent out; and determining a repair scheme of the intelligent building according to the type of the health abnormality. According to the method, the safety, comfort and management efficiency of the building can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of building health monitoring, and particularly to a health monitoring method and system for intelligent buildings. Background Art

[0002] With the rapid advancement of the urbanization process, intelligent buildings, as an important part of modern urban development, have increasingly entered people's daily lives. Intelligent buildings usually integrate information technology, automation control technology, sensing technology, and big data analysis technology, which can enhance the functionality, comfort, and energy efficiency of buildings. However, with the increase in the building's service life and the influence of external environmental changes, the health status of the building degenerates to varying degrees, resulting in problems such as potential safety hazards in the building structure, equipment failures, and unsuitable environmental conditions, thus affecting the building's usage effect, operation efficiency, and personnel safety.

[0003] Currently, the health monitoring of most buildings mainly relies on manual regular inspections and traditional monitoring equipment. This method has disadvantages such as untimely data acquisition, lagging information feedback, and untimely maintenance response, and it is difficult to achieve comprehensive, continuous, and accurate monitoring of the building's health status.

[0004] Therefore, developing a health monitoring method for intelligent buildings based on comprehensive data analysis and artificial intelligence optimization is of great significance for improving the intelligent level of building management, reducing building operation and maintenance costs, extending the building's service life, and ensuring personnel safety and comfort. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a health monitoring method and system for intelligent buildings to solve the above technical problems.

[0006] To achieve the above object, in a first aspect, a health monitoring method for an intelligent building is provided. The method includes the following steps: Real-time acquisition of multi-dimensional data inside and outside the intelligent building through multiple sensors set at multiple target monitoring positions in the intelligent building. The multi-dimensional data includes environmental data, structural data of the intelligent building, equipment status data, and personnel activity data; the structural data includes displacement, deformation, stress, vibration, and crack width of the intelligent building; Preprocess the multi-dimensional data to obtain preprocessed multi-dimensional data; According to the preprocessed multi-dimensional data, use a pre-trained health monitoring model to output an analysis result of the health status of the intelligent building. The health status analysis result includes the suitability of environmental conditions, the stability of the building structure, the health of equipment, and the risk level of personnel activities; When the health condition analysis result indicates that there is a health anomaly in the intelligent building, a warning signal is issued; According to the type of the health anomaly, determine a repair plan for the intelligent building, where the repair plan is at least associated with the repair and reinforcement of the building structure of the intelligent building.

[0007] In a second aspect, a health monitoring system for an intelligent building is provided. The system includes: A data acquisition module, which includes a plurality of sensors arranged at a plurality of target monitoring positions in the intelligent building, configured to acquire multi-dimensional data inside and outside the intelligent building in real time. The multi-dimensional data includes environmental data, structural data of the intelligent building, equipment status data, and personnel activity data; the structural data includes displacement, deformation, stress, vibration, and crack width of the intelligent building; A data preprocessing module, configured to preprocess the multi-dimensional data to obtain preprocessed multi-dimensional data; A health analysis module, configured to output a health condition analysis result of the intelligent building according to the preprocessed multi-dimensional data by using a pre-trained health monitoring model. The health condition analysis result includes the suitability of the environmental conditions, the stability of the building structure, the health of the equipment, and the risk level of personnel activities; A warning module, configured to issue a warning signal when the health condition analysis result indicates that there is a health anomaly in the intelligent building; A repair plan generation module, configured to determine a repair plan for the intelligent building according to the type of the health anomaly, where the repair plan is at least associated with the repair and reinforcement of the building structure of the intelligent building.

[0008] The above technical solution has the following beneficial technical effects: Through multi-dimensional data fusion and a closed-loop feedback mechanism, the intelligent building health monitoring method realizes real-time, comprehensive evaluation and dynamic optimization of the building health state. Through the collaborative monitoring of the environment, structure, equipment, and personnel activities, the accuracy and timeliness of anomaly detection are improved; the intelligent diagnosis based on the health monitoring model can accurately locate the root cause of the anomaly, and combined with the dynamic iteration of the adaptive repair plan, the fault response time is effectively shortened and the maintenance cost is reduced; at the same time, through data-driven warning and repair closed-loop management, not only the service life of the building is extended, but also the energy utilization efficiency and personnel safety protection ability are optimized, forming an intelligent operation and maintenance system with self-learning ability. Description of the Drawings

[0009] The drawings are used to better understand the present invention and do not constitute an improper limitation to the present invention. Among them: Figure 1 is a flowchart of the health monitoring method for an intelligent building according to an embodiment of the present invention; Figure 2 It is a mind map of the repair solution for the intelligent building in the embodiment of the present invention; Figure 3 It is a functional block diagram of a health monitoring system for the intelligent building in the embodiment of the present invention; Figure 4 It is another functional block diagram of the health monitoring system for the intelligent building in the embodiment of the present invention; Figure 5 It is a schematic structural diagram of the computer system in the embodiment of the present invention. Detailed implementation manners

[0010] The following makes an explanation of the exemplary embodiments of the present invention in conjunction with the accompanying drawings. Various details of the embodiments of the present invention are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted below. Embodiment 1

[0011] As Figure 1 shown, this embodiment provides a health monitoring method for an intelligent building. The method includes the following steps: S10: Real-time obtain multi-dimensional data inside and outside the intelligent building through multiple sensors arranged at multiple target monitoring positions of the intelligent building. The multi-dimensional data includes environmental data, structural data of the intelligent building, equipment status data, and personnel activity data.

[0012] Specifically, in this embodiment, various types of sensors are deployed in each room, corridor, hall and other areas inside the intelligent building to obtain multi-dimensional data in real time. For environmental data, temperature and humidity sensors, carbon dioxide concentration sensors, light intensity sensors and air quality sensors (such as sensors that can detect pollutants such as formaldehyde and PM2.5) are installed. These sensors are distributed in different floors and functional areas to collect environmental parameters such as temperature, humidity, carbon dioxide concentration, light intensity and the concentration of pollutants in the air. In terms of obtaining building structure data, pressure sensors, strain sensors, vibration sensors and displacement sensors are installed at key structural parts of the building, such as beam columns, walls, foundations, roofs, etc., to monitor structural data such as the stress condition, deformation degree, vibration amplitude and displacement change of the structure in real time. For equipment status data, current sensors, voltage sensors, temperature sensors and vibration sensors are installed on various types of equipment in the intelligent building, such as air conditioning systems, elevators, lighting equipment, water supply and power supply equipment, ventilation equipment, etc., to obtain equipment status parameters such as the operating current, voltage, working temperature and vibration frequency of the equipment. Regarding personnel activity data, cameras, infrared sensors and access control systems are deployed at public areas, entrances and exits, stairwells, etc. The cameras monitor the activity trajectories and gathering situations of personnel through image recognition technology. The infrared sensors detect the presence and movement of personnel. The access control system records the entry and exit times and identity information of personnel, so as to obtain personnel activity data. All sensors are connected to the data acquisition module through wired or wireless communication methods, and transmit the multi-dimensional data collected in real time to the data processing center.

[0013] S20: Preprocess the multi-dimensional data to obtain the preprocessed multi-dimensional data.

[0014] Specifically, in this embodiment, when preprocessing the multi-dimensional data obtained in step S10, data cleaning is first performed, and obvious abnormal data is removed by using preset rules and algorithms, such as data that exceeds the reasonable range due to sensor failures and data that appears errors during transmission. Then, data denoising processing is carried out, and corresponding denoising algorithms are adopted for different types of data. For example, for periodic noise in environmental data and equipment status data, Fourier transform is used for frequency domain filtering; for random noise in structural data and personnel activity data, median filtering or Gaussian filtering algorithms are adopted. Then, data normalization processing is carried out to convert data of different types and different dimensions into a unified numerical range. For example, the min-max normalization method is adopted to map the data to the [0, 1] interval for subsequent model processing. In addition, for missing data, according to the time series characteristics and correlation of the data, methods such as linear interpolation, polynomial interpolation or mean filling based on adjacent sensor data are used to complete the filling, ensuring that the preprocessed multi-dimensional data is complete, accurate and standardized.

[0015] S30: Based on the preprocessed multi-dimensional data, use the pre-trained health monitoring model to output the health condition analysis result of the intelligent building, where the health condition analysis result includes the suitability of the environmental conditions, the stability of the building structure, the health of the equipment, and the risk level of personnel activities.

[0016] Specifically, in this embodiment, after obtaining the preprocessed multi-dimensional data in step S20, the pre-trained health monitoring model is used for analysis. The health monitoring model is a neural network model based on deep learning, such as a combined model of a Convolutional Neural Network (CNN) and a Long Short-Term Memory Network (LSTM), which can effectively process multi-dimensional time series data. In the model training stage, a large amount of multi-dimensional data of historical intelligent buildings and the corresponding manually labeled health condition analysis results are collected as the training data set, and the model parameters are optimized through the backpropagation algorithm and the gradient descent method, so that the model can accurately learn the mapping relationship between the multi-dimensional data and the health condition. Input the preprocessed multi-dimensional data into the trained health monitoring model, and the health monitoring model outputs the suitability of the environmental conditions, the stability of the building structure, the health of the equipment, and the risk level of personnel activities respectively through feature extraction and non-linear transformation of the multi-layer neural network. Among them, the suitability of the environmental conditions is calculated according to the matching degree between environmental data such as temperature, humidity, and air quality and the human comfort interval and the normal operation interval of building equipment; the stability of the building structure is determined based on the comparative analysis of the structural data with the building design standards and historical normal state data; the health of the equipment is evaluated by the matching degree between the equipment status data and the normal operation parameter range and the fault characteristic pattern of the equipment; the risk level of personnel activities is judged according to whether there are abnormal gatherings, violations and other situations in the personnel activity data.

[0017] S40: When the health condition analysis result indicates that there is a health anomaly in the intelligent building, send out a warning signal.

[0018] Specifically, in this embodiment, when any one or more of the suitability of environmental conditions, the stability of building structures, the health of equipment, or the risk level of personnel activities in the health status analysis result output in step S30 exceed a preset threshold, indicating that there is a health anomaly in the intelligent building, the system will issue a warning signal. The warning signal includes various forms such as sound warning, light warning, SMS warning, and system interface prompt. For example, when the suitability of environmental conditions is lower than the safety threshold, a buzzer alarm sound is emitted in the corresponding area of the intelligent building, and a prompt is given through flashing lights; when the stability of the building structure shows an abnormal decline, in addition to on-site sound and light warnings, an SMS warning is also sent to the relevant personnel of the building management department to notify them to take measures immediately; when the health of the equipment detects that the equipment is about to malfunction, the abnormal information of the equipment is displayed on the monitoring system interface of the intelligent building, and a specific prompt sound is emitted; when the risk level of personnel activities is relatively high, for example, it is detected that the gathering of people may cause a safety accident, a safety prompt is issued through the broadcast system, and a warning message is displayed on the display screens in the relevant areas. The issuance of the warning signal can timely remind the relevant personnel to pay attention to the health anomalies of the intelligent building so as to quickly take countermeasures.

[0019] S50: Determine the repair plan for the intelligent building according to the type of the health anomaly, and the repair plan is at least associated with the repair and reinforcement of the intelligent building.

[0020] Specifically, in this embodiment, according to the type of health anomaly determined in step S40, the system will automatically determine the corresponding repair plan. If the type of health anomaly is unsuitable environmental conditions, such as too high or too low temperature, unqualified air quality, etc., the repair plan may include automatically adjusting the operating parameters of the air conditioning system, increasing the operating frequency of the ventilation equipment, or starting the air purification equipment, etc., to improve the environmental conditions. When the type of health anomaly is unstable or deformed building structure, the system will recommend immediately stopping the use of the relevant area, arranging professional building inspection personnel to conduct a detailed inspection and evaluation of the structure, formulating a structure reinforcement or repair plan according to the inspection results, and notifying the construction unit to carry out the construction. For equipment health anomalies, such as equipment malfunction or performance decline, the repair plan includes automatically switching to standby equipment, generating an equipment maintenance work order, and notifying the equipment maintenance personnel to conduct maintenance or replace parts. If the type of health anomaly is that there are risks or anomalies in personnel activities, such as potential safety hazards in the area where people gather, the repair plan includes guiding people to evacuate orderly through the broadcast, setting warning signs in the risk area, and increasing security personnel for on-site management. After the repair plan is determined, the system will send the plan to the relevant execution departments or personnel and track the repair process to ensure that the health anomalies of the intelligent building are handled in a timely and effective manner and restored to the normal operating state.

[0021] Such as Figure 2As shown, in some embodiments, when the abnormal type of the building structure is structural deformation, the following first repair plan associated with the repair and reinforcement of the intelligent building is determined: For slight deformation in the first deformation range (the deformation amount is within the allowable deviation range but close to the upper limit value), in combination with the original building structure design drawings, analyze the stress conditions of the deformed part. If the deformation is caused by local load concentration, the use function can be adjusted to reduce the load on this part. For example, move heavy equipment in this area. According to the construction specifications, check the structural components around the deformed part. If any connection looseness is found, re-tighten the connectors according to the specification requirements, such as tightening bolts, re-welding joints, etc. Adopt the reinforcement measure of pasting carbon fiber cloth. Cut the appropriate carbon fiber cloth according to the shape and size of the deformed part, treat the concrete surface according to the construction specification requirements, then apply structural adhesive, and paste the carbon fiber cloth on the surface of the deformed part to enhance the tensile capacity of the structure and limit the further development of the deformation.

[0022] For moderate deformation in the second deformation range (the deformation amount exceeds the allowable deviation range but does not affect the overall structural stability), formulate an unloading plan in combination with the building structure design and construction specifications. Set up a temporary support structure above the deformed part, such as steel supports or jacks, to transfer part of the upper load and reduce the stress on the deformed part. For concrete structures, the method of increasing the cross-section for reinforcement can be adopted. Determine the size and reinforcement of the newly added concrete according to structural calculations, roughen and clean the surface of the original structure, implant steel bars, and then formwork and pour concrete to make the newly added part work together with the original structure to improve the bearing capacity and stiffness of the structure. If it is the deformation of a steel structure, the method of welding stiffeners can be used for reinforcement. Weld stiffeners at the weak parts of the steel beam or steel column according to the deformation situation and design specification requirements to enhance the local stability and overall stiffness of the structure.

[0023] For severe deformation in the third deformation range (the deformation amount is large and affects the overall structural stability), immediately organize personnel to evacuate the site and set up obvious warning signs to prevent irrelevant personnel from entering the dangerous area. Develop a detailed overall reinforcement plan and adopt the method of external prestressing reinforcement. Arrange prestressing tendons outside the structure and apply a reverse acting force to the structure by tensioning the prestressing tendons to offset part of the deformation and improve the stress state of the structure. Demolish some severely deformed and irreparable structural components and rebuild them according to the original design and construction specification requirements. During the demolition and reconstruction process, ensure the protection of the surrounding structures to avoid further damage. The first deformation range, the second deformation range, and the third deformation range increase in sequence.

[0024] In some embodiments, when the abnormal type of the building structure is structural cracks, the following second repair plan associated with the repair and reinforcement of the intelligent building is determined: For microcracks within the first width range (crack width less than 0.05 mm), clean the cracks to remove dust and debris on the surface. According to the construction specifications, use the surface sealing method for treatment. Apply materials such as epoxy mortar to the crack surface to prevent the intrusion of moisture and harmful media and improve the durability of the structure.

[0025] For general cracks within the second width range (crack width between 0.05 mm and 0.3 mm), conduct pressure grouting treatment on the cracks. First, drill holes along both sides of the crack, and then use pressure grouting equipment to inject grouting materials such as epoxy resin into the crack to fill the crack voids and restore the integrity and waterproof performance of the structure. Inspect the structural members around the crack. If stress concentration is found due to reasons such as temperature change and concrete shrinkage, steel plates or carbon fiber sheets can be pasted on both sides of the crack for reinforcement to disperse stress and prevent the crack from further expanding.

[0026] For severe cracks within the third width range (crack width greater than 0.3 mm or the crack penetrates the structural member), first, detailed inspection and analysis are required to clarify the specific causes of the crack. The causes include foundation settlement, structural overload, concrete quality problems, construction defects, etc. After analyzing the causes of the crack, different treatment methods are adopted according to the specific situation. If the crack is caused by foundation settlement, the foundation needs to be reinforced. The reinforcement methods include measures such as bored piles and jacked piles to improve the bearing capacity of the foundation and prevent the crack from further expanding. If the crack is caused by structural overload, the stressed part needs to be redesigned or strengthened, such as by adding support structures or enhancing the bearing capacity of beams and columns. In addition, if the crack is caused by concrete quality problems or construction defects, the damaged part needs to be demolished and the concrete needs to be re-poured to restore the integrity of the structure. The repair measures should include cleaning and patching the crack. First, remove the loose materials inside the crack, such as dust and debris, and then select appropriate repair materials (such as epoxy resin and epoxy mortar) according to the width and depth of the crack for filling and sealing to prevent the intrusion of moisture and harmful substances. In addition, for larger cracks or insufficient structural bearing capacity, technologies such as steel plate reinforcement and carbon fiber sheet reinforcement can be used. By pasting steel plates or carbon fiber sheets on both sides of the crack, the strength and durability of the structure are enhanced, thereby restoring the integrity of the structure.

[0027] For through cracks, a method combining steel plate pasting and pressure grouting can be used for repair. First, conduct pressure grouting on the crack, and then paste steel plates on both sides of the crack. Connect the steel plates to the structural member firmly by bolts or welding to enhance the bearing capacity and integrity of the structure.

[0028] In some embodiments, when the abnormal type of the building structure is the deterioration of the performance of structural materials, determine the following third repair plan associated with the repair and reinforcement of the intelligent building: Regarding the deterioration of concrete materials (such as carbonation, steel corrosion, etc.), for mildly carbonated concrete structures, the method of applying concrete protective agents can be used to prevent the further intrusion of harmful gases such as carbon dioxide and delay the carbonation process. If steel corrosion is found, first chisel the concrete at the corroded part until the uncorroded steel bars are exposed. Remove the rust on the surface of the steel bars, for example, by using sandpaper grinding, chemical rust removal, etc., and then apply anti-rust paint. Repair the concrete at the chiseled part according to the requirements of the construction specifications. Polymer concrete or high-strength non-shrinking grouting material can be used for filling to ensure that the repaired concrete is closely combined with the original structure and restore the bearing capacity of the structure.

[0029] Regarding the deterioration of steel structure materials (such as steel corrosion, fatigue damage, etc.), for the rust on the surface of steel, it is treated according to the degree of rust. Mild rust can be removed by sandblasting to remove the rust layer on the surface of the steel, and then anti-corrosion coatings are applied. For severe rust, the severely corroded parts need to be cut and replaced, and the replaced steel should be treated for anti-corrosion. If there is fatigue damage in the steel structure, the anti-fatigue ability of the structure can be improved by increasing the redundancy of the structure or adopting reinforcement measures. For example, welding steel plates or pasting carbon fiber cloth in the tensile area of the steel beam to share part of the stress and reduce the influence of fatigue damage. Regularly detect and maintain the steel structure, and check and tighten the connection parts of the structure according to the requirements of the design and construction specifications to ensure the safety and reliability of the structure.

[0030] In the health monitoring of intelligent buildings, total stations, levels and other measuring instruments are used for coordinate and elevation measurement in structural deformation detection, a three-dimensional model is obtained by means of a laser scanner, and dynamic monitoring is realized by using strain and displacement sensors at the same time; in structural crack detection, a crack width gauge is used to measure the width, and internal cracks are detected by non-destructive equipment such as ultrasonic flaw detectors; in the detection of the deterioration of structural material properties, for concrete, rebound hammers, carbonation depth gauges, core sampling and other methods are used to detect strength, carbonation degree and mechanical properties, while for steel structures, magnetic particle flaw detectors, ultrasonic thickness gauges, chemical analysis and fatigue test equipment are used to detect defects, thickness, chemical composition and fatigue performance.

[0031] The above technical solution collects environmental, structural, equipment status, and personnel activity data of intelligent buildings in real time through multi-dimensional sensors. After preprocessing such as data cleaning, noise reduction, normalization, and missing data filling, the integrity and reliability of the data are ensured. The multi-dimensional data is intelligently analyzed using deep learning models, and quantitative evaluation results of the environmental suitability level, structural stability level, equipment health level, and personnel activity risk level can be accurately output, realizing a systematic diagnosis of the building's health status. Through diverse warning methods such as sound and light, text messages, and system interfaces, the abnormal response mechanism can be triggered in real time. Combined with the automatically generated differentiated repair solutions (such as equipment parameter adjustment, structural inspection and repair, personnel guidance and management, etc.), a closed-loop management system is formed. This solution effectively improves the comprehensiveness, real-time nature, and accuracy of intelligent building health monitoring, can identify environmental safety hazards, structural performance degradation, equipment failure risks, and personnel gathering risks in advance, provides intelligent decision-making support for the safe operation of buildings, efficient maintenance of equipment, comfortable regulation of the environment, and safety management of personnel, reduces the cost of manual inspections, and enhances the reliability and emergency response capabilities of building systems. Embodiment 2

[0032] In this embodiment, according to the intelligent building health monitoring method described above, the step S10 includes: S101: Real-time collect the environmental data inside the intelligent building through environmental sensors, and the environmental data includes temperature, humidity, air quality, and light intensity.

[0033] In this embodiment, the collection of environmental data inside the intelligent building is realized through distributed environmental sensors. Specifically, in areas such as rooms, corridors, halls, and equipment rooms on each floor, temperature and humidity sensors are installed at intervals of every 5 - 10 square meters to collect temperature (accuracy ±0.3°C) and relative humidity (accuracy ±2% RH) in real time; air quality sensors (such as a combination of SPS30 type particulate matter sensors and CCS811 type gas sensors) are deployed at the ceiling or ventilation openings to detect PM2.5, PM10 concentrations, and volatile organic compound content to characterize air quality; light intensity sensors (such as BH1750 type digital sensors) are installed near windows and in artificial lighting areas to monitor the light intensity of natural light and artificial light sources (unit: lux) in real time. All environmental sensors are connected to the data collection module through Modbus or ZigBee communication protocols to ensure that data is collected and uploaded to the central server at least once per minute.

[0034] S102: Real-time collect the structural data of the intelligent building through structural health monitoring sensors, and the structural data includes the displacement, deformation, stress, vibration, and crack width of the intelligent building.

[0035] Specifically, for the acquisition of intelligent building structure data, structural health monitoring sensors are deployed at key load-bearing parts and vulnerable nodes of the building. Among them, resistance strain gauges are pasted at positions such as beam-column joints, the bottom of shear walls, and foundation caps, and cooperate with a static strain gauge to monitor the structural stress change in real time (accuracy ±0.1% FS); laser displacement sensors are installed on the roof, floor cantilever parts, and the inner wall of the elevator shaft, and the displacement deformation of structural members is measured by emitting laser beams (accuracy ±1μm); acceleration vibration sensors are fixed on the surface of each floor slab and the core wall to collect structural vibration signals at a sampling frequency of 100Hz for analyzing parameters such as vibration frequency and amplitude; for masonry walls or concrete structures prone to cracks, crack width sensors are installed at crack monitoring points to measure the change in crack width through the deformation of precision resistance wires (resolution 0.01mm). The above sensors transmit data to the structural health monitoring subsystem through wired cables (shielded twisted pairs) or LoRa wireless modules.

[0036] S103: Real-time collect the device status data of the intelligent building through device status monitoring sensors, where the device status data includes the operating status data of the power system, heating system, ventilation system, and air conditioning system.

[0037] Specifically, the acquisition of device status data is achieved through dedicated sensors integrated into each system device: current transformers and voltage sensors are installed on transformers, distribution cabinets, and transmission lines in the power system to monitor parameters such as three-phase voltage, current, and active power in real time; thermocouple temperature sensors and pressure sensors are deployed on the surface of boilers, pipelines, and heat exchangers in the heating system to collect supply water temperature, return water pressure, and flow rate data; wind speed sensors and differential pressure sensors are installed at the positions of fans, air ducts, and air vents in the ventilation system to monitor the ventilation volume and pipeline pressure difference; temperature sensors, humidity sensors, and electronic expansion valve opening sensors are installed at the compressor, condenser, and terminal air vents in the air conditioning system to obtain the supply air temperature, refrigeration power, heating power, and device operating current in real time. All device sensors are connected to the device management system through industrial Ethernet (Modbus TCP protocol) or 485 bus to achieve data update at the second level.

[0038] S104: Real-time collect the personnel activity data inside the intelligent building through a personnel positioning and monitoring device, where the personnel activity data includes the distribution density of personnel, the activity trajectories of personnel, and the residence duration of personnel in different areas.

[0039] Specifically, the collection of personnel activity data relies on a multi-modal fusion personnel positioning and monitoring device. UWB positioning base stations are deployed on the ceilings of various areas in the intelligent building, cooperating with the positioning tags worn by personnel (integrated with accelerometers and gyroscopes), to achieve real-time positioning with centimeter-level accuracy, obtaining personnel coordinates and activity trajectories; infrared pair sensors are installed in corridors, stairwells and other passageways, and a counting module is used to count the frequency of personnel flow between areas; binocular cameras are installed at main entrances and exits and in elevators, combined with deep learning object detection algorithms (such as YOLOv8), to identify the number of personnel, distribution density and residence duration (implemented through trajectory tracking algorithms); pressure-sensitive floor tiles (such as a thin-film pressure sensor array) are laid on the ground in areas where people gather, such as meeting rooms and restaurants, to assist in verifying personnel density data. All monitoring devices transmit data to the personnel management server through Wi-Fi or 5G networks, and after privacy protection processing, the data is stored in the database.

[0040] The advantages of this embodiment are that through the combined application of multiple sensors, comprehensive and real-time monitoring of all aspects of the intelligent building is achieved. Environmental sensors provide accurate data such as temperature and humidity, which is beneficial to ensuring the suitability of the internal environment of the building; structural health monitoring sensors can detect the stability of the building structure in real time and timely discover possible risks; equipment status monitoring sensors ensure the efficient operation of various equipment in the building and avoid equipment failures; while the collection of personnel activity data helps to optimize the use efficiency and safety management of the building space. This method can provide a full range of accurate health status assessment and early warning for the intelligent building, effectively improving the safety, comfort and energy efficiency of the building. Embodiment 3

[0041] In this embodiment, in the health monitoring method of the intelligent building, the step S20 includes: S201: Denoise the multi-dimensional data to obtain purified data.

[0042] Specifically, in the denoising process of step S201, different filtering algorithms are adopted for different types of multi-dimensional data. For environmental data (data with periodic characteristics such as temperature and humidity), wavelet transform denoising is used: first, multi-layer wavelet decomposition is performed on the time series data (for example, db4 wavelet basis is selected and the decomposition level is set to 3 layers), high-frequency noise components are identified and attenuated in the frequency domain, and then the purified signal is reconstructed through inverse wavelet transform; for the vibration acceleration signal in the structural data, Kalman filtering is used for dynamic denoising. By establishing a state space model of structural vibration (the state vector includes displacement, velocity, and acceleration), the state estimate is recursively updated using the residual between the sensor measurement value and the predicted value, effectively filtering out environmental vibration interference; for the impulse noise in the equipment status data (such as current instantaneous spikes), median filtering (the window size is set to 5 sampling points) is used to remove isolated noise points; for the positioning trajectory jitter problem in the personnel activity data, the moving average filtering (the window size is set to 30 seconds) is used to smooth the trajectory coordinates to ensure the continuity of the position data. All denoising algorithms are implemented through a data processing engine (such as the Scipy library in Python) and support multi-threaded parallel processing.

[0043] S202: Perform anomaly detection on the purified data, remove the abnormal data points, and obtain the cleaned data.

[0044] Specifically, in step S202, the detection of abnormal data points is realized by combining statistical methods and machine learning algorithms. First, for numerical data (such as stress, voltage, PM2.5 concentration), the IQR (Interquartile Range) method is used to set dynamic thresholds: calculate the 25th percentile (Q1) and the 75th percentile (Q3) of the data, and define the outliers as the data points less than Q1 - 1.5IQR or greater than Q3 + 1.5IQR, and automatically mark and remove them; for time series data (such as structural displacement, equipment operating power), the Isolation Forest algorithm is used to detect local anomalies. By constructing a random binary tree, the isolation degree of data points is quantified, and the data points with an isolation score exceeding 0.8 are determined as anomalies; for spatially correlated data (such as the correlation between personnel density and regional area), the spatial distance clustering algorithm (DBSCAN) is used to identify outliers. The neighborhood radius is set to 2 meters and the minimum sample number is set to 5 people, and the abnormal aggregated data deviating from the main clustering cluster is corrected. During the anomaly detection process, the system automatically generates an anomaly data report, recording the anomaly type, occurrence time, and associated sensor number for model optimization.

[0045] S203: Perform standardization processing on the cleaned data to obtain the standardized data.

[0046] Specifically, in step S203, according to the input requirements of the subsequent health monitoring model, the cleaned data is uniformly converted into a standardized numerical range. For a neural network model (such as the CNN-LSTM model in step S30), min-max normalization is adopted to map the data to the interval [0, 1]; for distance-based models such as Support Vector Machine (SVM), Z-score normalization is used to make the data follow a standard normal distribution. For multi-source heterogeneous data (for example, the unit of temperature is °C, the unit of stress is MPa, and the unit of light intensity is lx), the system automatically identifies the data dimension and calls the corresponding standardization module to support the online standardization of real-time data streams, ensuring that the standardized data is compatible with the input requirements of multiple models.

[0047] S204: Perform time series analysis on the standardized data to obtain preprocessed data.

[0048] Specifically, in the time series analysis stage of step S204, first, the standardized data is sorted by timestamp to generate a time series matrix with equal time intervals (such as 1 minute). For data with missing timestamps, cubic spline interpolation is used for completion to ensure the continuity of the time series; then, through the sliding window technique (the window size is set to 1 hour and the step size is set to 10 minutes), time series features are extracted to generate input samples containing the data at the current moment and historical context information (for example, each sample contains multi-dimensional data of the previous 6 time points); for environmental data with strong periodicity (such as the diurnal variation of temperature and humidity), Fourier transform is used to extract frequency domain features to identify the periodic components in the data (such as 24-hour period, 7-day period), and they are spliced with the original time domain data to form a composite feature vector; for the stay duration and trajectory data in human activity data, time encoding techniques (such as converting the stay duration to the time proportion relative to a day) are incorporated into the time series model. The time series analysis module is developed based on the TensorFlow or PyTorch framework, supporting the generation of preprocessed data that conforms to the input format of time series models such as LSTM and Transformer, and at the same time outputting a heat map of data time correlation for adjusting feature weights during model training. Embodiment 4

[0049] In this embodiment, in the health monitoring method of the intelligent building, the step S30 includes: S301: Input the preprocessed multi-dimensional data into the health monitoring model, where the health monitoring model includes a data fusion layer, a feature extraction layer, and a health assessment layer.

[0050] Specifically, after completing the multi-dimensional data preprocessing, the preprocessed multi-dimensional data is input into the health monitoring model in the form of a time series matrix. The model is designed with a hierarchical architecture, including a data fusion layer, a feature extraction layer, and a health assessment layer, and is implemented based on the TensorFlow or PyTorch framework. The input data format is a three-dimensional tensor [number of samples, time step, data dimension], where the time step corresponds to the unified time interval in the preprocessing stage (e.g., 1 minute), and the data dimension integrates the standardized features of four types of data: environment, structure, equipment, and personnel activities (e.g., temperature, stress, equipment current, personnel density, etc., a total of 50 dimensions). The model input interface supports dynamic data batch loading, and realizes the asynchronous processing of preprocessed data and model calculation through a data queue, improving the real-time monitoring efficiency.

[0051] S302: The data fusion layer performs spatio-temporal alignment and dimensional normalization processing on the multi-dimensional data to generate a standardized data set.

[0052] Specifically, the data fusion layer first performs spatio-temporal alignment processing, which includes: for the time dimension, through a resampling algorithm, the original sampling frequencies of different sensors (e.g., 100Hz for the structural vibration sensor and 1Hz for the environmental sensor) are unified to the time step required by the model input (e.g., 1 minute), the mean downsampling is used for high-frequency data, and cubic spline interpolation upsampling is used for low-frequency data; for the space dimension, a three-dimensional space coordinate mapping table of the intelligent building is established, and the positions of structural sensors (e.g., beam-column coordinates), equipment installation positions (e.g., air conditioner unit numbers), and personnel positioning coordinates (UWB positioning tag coordinates) are uniformly converted into the Building Information Modeling (BIM) coordinate system to form a space feature vector containing X / Y / Z coordinates. Secondly, dimensional normalization processing is performed: for cross-type data (e.g., the unit of temperature in °C and the unit of stress in MPa), standardized scaling is used. By calculating the global mean and standard deviation of each dimension data (statistical based on historical 30-day data), the data is converted into a standard normal distribution with a mean of 0 and a standard deviation of 1, generating a standardized data set with unified dimensions [N, T, D] (N is the number of samples, T is the time step, and D is the dimension of the fused data).

[0053] S303: The feature extraction layer extracts environmental features, structural features, equipment features, and personnel activity features from the standardized data set through a convolutional neural network or a graph neural network.

[0054] Specifically, the feature extraction layer processes by branching paths according to the data type differences: In the step of extracting environmental features and equipment features, a 1D convolutional neural network (1D-CNN) is adopted. For time series data (such as time series of temperature and humidity, equipment current), three convolutional layers are set, and the kernel sizes of the convolutional layers are 15, 10, and 5 respectively (corresponding to time windows of 15 minutes, 10 minutes, and 5 minutes). The ReLU (Rectified Linear Unit) activation function is used to extract periodic features at different time scales, and environmental feature vectors and equipment feature vectors with an output dimension of [N, T, 64] are obtained.

[0055] In the step of extracting structural features, a graph neural network (Graph Neural Network, GNN) is adopted. The key nodes of the building (such as beams, columns, shear walls) are defined as graph nodes (the number of nodes M = 200 - 500), and the node attributes include structural data such as stress, displacement, and vibration. The edge weights between nodes are set according to the physical connection relationship (such as the stiffness matrix of adjacent beams and columns). Two layers of graph convolutional network (Graph Convolutional Network, GCN) are used to aggregate the features of neighboring nodes, and structural coupling features (such as the correlation between node displacement and the stress of adjacent beams and columns) are extracted, and a structural feature vector with an output dimension of [M, 128] is obtained.

[0056] In the step of extracting human activity features, a spatio-temporal graph network (Spatio-Temporal Graph Neural Network, ST-GNN) is adopted. Each area of the building is defined as a graph node (the number of nodes K = 50 - 100), and the node attributes include the density of personnel distribution and the staying duration. The edge weights are set according to the regional connectivity (such as rooms connected by corridors). The LSTM layer in the time dimension is combined to capture the temporal law of personnel flow, and a human activity feature vector with an output dimension of [K, 96] is obtained. The feature extraction results of each branch are integrated into a comprehensive feature tensor [N, T, 288] (64 + 128 + 96) through a concatenation operation and input into the health assessment layer.

[0057] S304: The health assessment layer respectively outputs the environmental condition suitability score, the building structure stability index, the equipment health percentage, and the human activity risk level according to the environmental features, the structural features, the equipment features, and the human activity features.

[0058] Specifically, the health assessment layer includes four independent assessment sub-modules, which respectively output quantization results.

[0059] In the environmental condition suitability evaluation sub-module, a fully connected neural network (with 3 layers and the number of neurons being 256, 128, and 1 respectively) is adopted. The environmental feature vector is input, and a score ranging from 0 to 100 is output (above 80 points is suitable, 60 - 80 points is critical, and below 60 points is unsuitable). During training, the human comfort interval (such as temperature 22 - 26°C, humidity 40% - 60%) and the suitable operating interval of the equipment are used as labels, and the mean squared error (MSE) is adopted as the loss function.

[0060] In the building structure stability index sub-module, based on the structural feature vector, a dimensionless index ranging from 0 to 1 is calculated through a risk prediction model (support vector regression SVR) (above 0.9 is stable, 0.7 - 0.9 is a warning, and below 0.7 is dangerous). When training the model, the standard stress threshold of the building design and the historical crack propagation data are fused as constraints. In the equipment health percentage sub-module, for the equipment feature vector, the attention mechanism (Attention) is adopted to focus on key fault features (such as abnormal current fluctuations, vibration frequency offsets). The probability distribution of the equipment health state (healthy, sub-healthy, pre-fault, fault) is output through a Softmax classifier, and it is converted into a health percentage ranging from 0 to 100% (healthy corresponds to 90% - 100%, and fault corresponds to 0 - 40%).

[0061] In the personnel activity risk level sub-module, a risk assessment decision tree is constructed. The personnel activity features (density, trajectory, stay duration) are input, and three-level thresholds (low risk, medium risk, high risk) are set. For example, when the personnel density in the area is greater than 1.5 people / ㎡ and the stay duration is greater than 30 minutes, it is determined as medium risk. The decision tree branch conditions are optimized by combining historical accident data, and the discrete risk level (level 1 to level 3) is output. The output results of each sub-module are synchronized to the visualization interface, supporting real-time dynamic rendering of the health state and triggering of threshold alarms.

[0062] The advantages of the above technical solution are as follows. Through the collaborative work of the data fusion layer, feature extraction layer, and health assessment layer in the health monitoring model, the efficient integration and in-depth analysis of multi-dimensional data are realized. The data fusion layer ensures the unity and accuracy of the data through spatio-temporal alignment and dimensional normalization processing; the feature extraction layer adopts a convolutional neural network or a graph neural network, which can automatically extract key features from the standardized dataset to ensure comprehensive coverage of environmental, structural, equipment, and personnel activity features; the health assessment layer outputs accurate health assessment results respectively based on these features, such as environmental suitability, structural stability, equipment health percentage, and personnel activity risk level.

[0063] In some embodiments, the output of the environmental condition suitability score in step S304 includes: inputting the environmental features into a pre-constructed environmental suitability assessment sub-model, which is trained based on a support vector machine or a random forest algorithm. The environmental features at least include temperature and humidity data, air quality index, light intensity, and noise decibel value. The environmental suitability assessment sub-model classifies and quantifies the environmental features through a preset environmental standard threshold matrix, generating an environmental condition suitability score ranging from 0 to 10 points.

[0064] In some embodiments, the output of the building structure stability index in step S304 includes: inputting the structural features into a structural health monitoring sub-model. The structural features at least include building settlement data, wall crack width, beam-column stress value, and foundation vibration frequency. The structural health monitoring sub-model calculates the difference between the finite element analysis model and the real-time monitoring data, and combines a preset structural safety level knowledge base to calculate a building structure stability index ranging from 0 to 1, where the value closer to 1 indicates higher stability.

[0065] In some embodiments, the output of the equipment health percentage in step S304 includes: inputting the equipment features into an equipment fault prediction sub-model. The equipment features at least include elevator operation parameters, air-conditioning energy consumption data, power supply system voltage fluctuation value, and water supply and drainage pipeline pressure value. The equipment fault prediction sub-model is based on a Long Short-Term Memory Network (LSTM) or an attention mechanism model, performs time series analysis on the equipment features, and generates an equipment health percentage ranging from 0% to 100% by calculating the deviation degree between the current state and the baseline of the normal operation state of the equipment.

[0066] In some embodiments, the output of the personnel activity risk level in step S304 includes: inputting the personnel activity features into a personnel safety assessment sub-model. The personnel activity features at least include personnel density distribution, emergency exit occupancy, fire equipment usage status, and abnormal behavior recognition data. The personnel safety assessment sub-model performs risk factor weighted calculation on the personnel activity features based on a combination of a rule engine and machine learning, generating a three-level personnel activity risk level including low risk, medium risk, and high risk.

[0067] In some embodiments, step S30 further includes: before outputting the health status analysis result, performing a comprehensive weighted calculation on the environmental condition suitability, building structure stability, equipment health, and personnel activity risk through a preset multi-dimensional weight matrix, generating a building health comprehensive index ranging from 0 to 100. The multi-dimensional weight matrix dynamically adjusts the weight coefficients of each dimension according to the usage type of the intelligent building (residential, commercial, industrial).

[0068] In some embodiments, the pre-trained health monitoring model in step S30 is obtained through the following steps: collecting historical multi-dimensional data as a training data set, where the historical multi-dimensional data includes normal operation data, single fault data, and compound fault data; performing data augmentation and feature engineering processing on the training data set to generate a standard training set with health labels; using a multi-task learning framework to train the health monitoring model, and simultaneously optimizing the loss functions of four sub-tasks of environmental suitability, structural stability, equipment health, and personnel activity risk, where the loss function includes a combination of mean square error loss and cross-entropy loss.

[0069] In some embodiments, it further includes: introducing an adversarial sample training mechanism during the training process, and performing robustness training on the health monitoring model by generating adversarial samples that simulate abnormal data, so as to improve the model's ability to identify sudden faults and data noise. Embodiment Five

[0070] In this embodiment, in the health monitoring method of the intelligent building, step S40 includes: S401: According to the health condition analysis result, determine the health abnormal type of the intelligent building, where the health abnormal type includes unsuitable environmental conditions, abnormal or unstable building structure, equipment failure, and abnormal personnel activities.

[0071] Specifically, after the health status analysis result is output in step S30, the system automatically matches the health abnormality type through a preset determination rule. For unsuitable environmental conditions, it is determined as an environmental abnormality when the environmental condition suitability score is lower than 60 points and continuously lower than this threshold for at least 3 time steps (such as 3 minutes). Specifically, it is subdivided into temperature abnormality (less than 18°C or greater than 28°C), humidity abnormality (less than 30% RH or greater than 70% RH), air quality exceeding the standard (PM2.5 greater than 75 μg / m³ or volatile organic compounds greater than 500 ppb), or light abnormality (less than 300 lux or greater than 3000 lux). The determination condition for building structure abnormality is that the building structure stability index is lower than 0.7, or any one of the monitoring values of structural displacement, stress, and crack width exceeds the design safety threshold (such as displacement greater than 5 mm, stress greater than 80% of the design value, crack width greater than 0.3 mm), and is triggered in association with abnormal vibration frequency (such as non-load vibration above 10 Hz). The determination basis for equipment failure is that the equipment health percentage is lower than 40%, or the equipment status data (such as current, voltage, vibration) shows a mutation (exceeding 3 times the standard deviation of the mean) and lasts for 2 time steps (such as 2 minutes), and in combination with the historical data of the same type of failure in the equipment operation log, it is subdivided into power system trip warning, heating pipeline pressure abnormality, air conditioner compressor overheating, etc. The determination standard for abnormal personnel activities is that the personnel activity risk level reaches level 3 (high risk), or specific patterns such as the monitored personnel density being greater than 2 people / m² and the stay duration being greater than 45 minutes, reverse congestion in the activity trajectory (such as people flowing backwards in the stairwell), and unauthorized area intrusion are detected. The abnormal type determination rule is stored in the system configuration database, supporting the administrator to customize the threshold and association logic through the visual interface.

[0072] S402: For each health abnormality type, set corresponding warning trigger conditions, and the warning trigger conditions include that the health status exceeds a preset threshold.

[0073] Specifically, the system presets multi-level warning trigger conditions for each type of health anomaly and supports dynamic configuration. For the case of unsuitable environmental conditions, the system sets three-level warning trigger conditions: at the primary warning level (60 to 80 points), area-level prompts are triggered, such as displaying the current environmental parameters through the LED screen on each floor; at the intermediate warning level (40 to 60 points), audible and visual alarms are triggered, using a buzzer and yellow lights for prompting; at the advanced warning level (score below 40 points), global warnings are triggered and device adjustments are linked, such as automatically starting the fresh air system. For abnormal or unstable building structures, a composite trigger mechanism of "single parameter overrun and associated parameter trend" is adopted. For example, when the stress of a certain beam-column exceeds 70% of the design value and the displacement growth rate of adjacent nodes exceeds 0.5 mm / h, an orange warning is triggered; if the stress exceeds 90% of the design value or the daily growth rate of the crack width exceeds 0.1 mm, a red warning is triggered. The trigger conditions for equipment failures are based on the matching degree between the equipment health percentage and the fault feature library. When the equipment health is between 30% and 40%, a "pre-fault" warning is triggered to prompt the maintenance plan; when the health is below 30% or hard fault features such as sudden current drop and vibration peak appear, an "emergency shutdown" warning is triggered. For abnormal human activities, the system dynamically adjusts the trigger conditions according to the risk level and real-time passenger flow. During the off-peak period (less than 50 people per floor in passenger flow), if the personnel density exceeds 1.5 people / ㎡, a medium-risk warning is triggered; during the peak period (more than 200 people per floor in passenger flow), if the personnel density exceeds 2.0 people / ㎡ or there is aggregation and stay (stay duration exceeds 60 minutes), a high-risk warning is triggered. All trigger conditions include time window constraints (such as an anomaly lasting for 5 minutes before it can be triggered). These rules use a rule engine (such as Drools) to achieve efficient matching of conditional logic, thus ensuring that the system can respond to various health anomaly events in a timely and accurate manner. S403: Send a warning signal according to the warning trigger conditions.

[0074] Specifically, the early warning signal sending module supports differentiated early warning strategies for multiple channels, multiple levels, and multiple objects. The early warning channels include local audible and visual alarms (warning lights and buzzers installed in each area of the building, supporting zonal control), remote notifications (sms / wechat push to the administrator's mobile phone, email sent to the operation and maintenance team, APP pop-up reminder for the on-duty personnel), and system-level linkages (sending control commands to the building automation system, such as cutting off the power supply of faulty equipment; highlighting the abnormal location and marking the risk level on the BIM visualization interface). The signal content includes the abnormal type (such as "air conditioning system failure on the 3rd floor"), specific parameters (such as "current current 120A, rated current 80A"), occurrence location (accurate to BIM coordinate system X = 15.2m, Y = 8.5m, Z = 3.8m), duration, number of historical similar abnormalities, and recommended response measures (such as "please immediately check the ventilation duct in Room 302"). In terms of the priority mechanism, the red early warning (structural instability / urgent equipment failure) adopts a triple confirmation mechanism (sms, phone voice, on-site broadcast), the orange early warning (advanced environmental abnormality / high risk of personnel) triggers high-frequency push, the yellow early warning (primary abnormality) only flashes a reminder on the monitoring interface, and the early warning signal is asynchronously distributed through a message queue (such as RabbitMQ) to ensure reliable transmission in high-concurrency scenarios. At the same time, an early warning log is generated and stored in the time series database (InfluxDB) to record the early warning time, processing status, and closed-loop result for subsequent fault analysis and model optimization.

[0075] In some embodiments, after step S40, there is also step S70: when an environmental early warning signal is triggered, the ventilation and purification equipment is linked to be turned on and the temperature and humidity control system is adjusted; when a structural early warning signal is triggered, the entrances and exits of the abnormal area are automatically locked and heavy equipment operation is prohibited; when an equipment early warning signal is triggered, the standby equipment is switched to and a maintenance work order is generated and pushed to the maintenance unit; when a personnel early warning signal is triggered, the emergency broadcast system is started and the emergency evacuation indicator lights are lit.

[0076] Specifically, when the environmental condition suitability score is lower than the preset threshold and the early warning type is determined to be environmental (such as abnormal temperature and humidity, excessive air quality), an instruction is automatically sent to the ventilation and air conditioning control system to turn on the fresh air purification equipment and adjust the temperature and humidity control parameters (such as adjusting the humidity control target value to the range of 40%-60%), and at the same time, a linkage signal is sent to the intelligent lighting system to dynamically adjust the supplementary lighting strategy according to the real-time illumination intensity.

[0077] When the building structure stability index is lower than the safety threshold (e.g., 0.6) or when it is detected that the width of the wall crack exceeds the limit, the system immediately sends instructions to the access control management system to lock the entrances and exits of the abnormal structure area (e.g., floor evacuation doors, equipment room passages), prohibits the operation of heavy equipment (e.g., elevators, goods lifts) in this area through the building automation system, and at the same time marks the abnormal structure position in real-time in the BIM model and pushes it to the structure engineer's terminal.

[0078] If the equipment health percentage is lower than the critical value (e.g., 70%) and it is determined as an equipment failure warning (e.g., abnormal elevator operation parameters, over-limit voltage fluctuation in the power supply system), the system first triggers the automatic switching mechanism of the standby equipment (e.g., switching of the dual-power system, redundant startup of the air conditioning unit), and then generates an electronic maintenance work order containing the equipment model, fault code, and historical maintenance records through the work order management module, pushes it to the management system of the equipment maintenance unit through the application programming interface, and attaches a 3D coordinate map of the equipment installation location.

[0079] When the risk level of personnel activities is determined as "high risk" (e.g., excessive personnel density, blocked emergency passage), the system immediately starts the emergency response procedure: plays the pre-recorded emergency evacuation voice through the broadcast control system, synchronously lights up all the emergency evacuation indicators in the building and dynamically marks the optimal escape route in the digital twin model; sends a pre-start instruction to the fire control system (e.g., preheating of the fire pump, standby of the smoke exhaust fan), and at the same time encrypts and uploads the real-time personnel distribution heat map and camera monitoring images to the urban emergency management platform.

[0080] The above linkage control steps realize the data interaction of different subsystems through a preset device linkage protocol (e.g., Modbus / TCP, BACnet). All linkage instructions are attached with timestamps and operation logs, and are stored through blockchain technology to ensure the traceability and operation standardization of the emergency response process. Embodiment Six

[0081] In this embodiment, in the health monitoring method of the intelligent building, the step S50 includes: S501: According to the type of the health abnormality, identify the abnormal target object corresponding to the health abnormality, and the abnormal target object includes the ventilation system, load-bearing structure, elevator equipment or personnel-intensive area of the intelligent building. Specifically, after determining the type of health abnormality in step S41, the system relies on a preset mapping relationship table and a Building Information Model (BIM) to automatically identify abnormal target objects through the association algorithm between sensor spatial coordinates and physical entities. When the type of health abnormality is that the environmental conditions are unsuitable, if the monitored temperature or humidity value exceeds the comfort range, the system locates the corresponding air conditioner unit, humidifier or dehumidification equipment according to the deployment location of the sensor corresponding to the abnormal data (such as the temperature and humidity sensor at the air outlet of the air conditioner on a certain floor); if the air quality parameter (such as PM2.5) exceeds the standard, it is associated with the fresh air system, air purification equipment or ventilation duct. For the situation of unstable building structure, the system accurately locks the specific load-bearing structural components (such as the frame beam on the 3rd floor and the bottom node of the shear wall in the basement 1st floor) by combining the coordinate data of the structural sensors (such as beam and column strain gauges, crack sensors) with the three-dimensional spatial positioning of the BIM model. For equipment failures, the system matches the equipment ledger database through the unique ID of the equipment status sensor (such as the elevator traction machine vibration sensor code, the current transformer number of the power distribution cabinet) and directly locates the faulty equipment (such as the 5th elevator control cabinet, the 2nd floor central air-conditioning compressor). When abnormal personnel activities occur, the system determines the specific area where people gather based on the coordinate data of the Ultra-Wideband (UWB) positioning tags or the camera video analysis results, and associates the access control system, monitoring equipment and fire-fighting facilities in this area as auxiliary target objects to ensure that the physical entity of the repair plan is clearly pointed to.

[0082] S502: Retrieve at least one basic repair strategy that matches the type of the health abnormality and the abnormal target object from a preset repair strategy database, where the repair strategy database stores basic repair strategies for different types of health abnormalities and the abnormal target objects.

[0083] Specifically, the repair strategy database is built using a relational database (such as PostgreSQL). Its storage structure is designed as a five-tuple data model consisting of anomaly type, target object, basic repair strategy, execution priority, and safety constraints. For example, for the anomaly combination of "unsuitable environmental conditions - ventilation system," the basic strategies stored in the database include: when PM2.5 concentration exceeds 150μg / m³, a control instruction automatically triggers the activation of the fresh air system to maximum air volume and the closing of the return air valve; when a volatile organic compound leak is detected, a safety strategy is implemented to shut down the ventilation ducts in the contaminated area and activate air purification equipment. For the anomaly scenario of "unstable building structure - load-bearing beams and columns," if the stress in the beams and columns reaches 70% of the design value and the displacement growth rate exceeds 0.5mm / h, the system initiates an early warning strategy by deploying drone infrared scanning of the structure surface and deploying temporary monitoring points. If the stress exceeds 90% of the design value or the crack width increases by more than 0.1mm per day, an emergency strategy is implemented to initiate temporary structural support and reinforcement procedures and prohibit entry to the area. For "equipment failure - elevator equipment", when the equipment health is at the pre-failure stage of 30%-40%, the system generates a maintenance strategy that triggers the operation of the backup elevator and pushes a maintenance work order to the maintenance team; if hard fault characteristics such as current drop or vibration peak are detected, the safety strategy of immediately stopping the equipment and cutting off the power supply is called. For "abnormal personnel activities - crowded areas", when the personnel density exceeds 1.5 people / ㎡ during off-peak hours, a medium-risk strategy of prompting orderly evacuation through the broadcast system is implemented; when the density exceeds 2.0 people / ㎡ during peak hours or the gathering stay time exceeds 60 minutes, a high-risk strategy of linking access control to restrict entry to the area and notifying security personnel for on-site control is activated. When the policy is called, the system uses SQL query language to match the combination of the anomaly type and the target object, and automatically retrieves the latest version of the policy, supporting historical policy version tracing and security compliance verification.

[0084] S503: Optimize the retrieved basic repair strategy based on the pre-processed multi-dimensional data and historical repair records to generate a repair plan for the intelligent building.

[0085] Specifically, the optimization process of the basic repair strategy combines real-time multi-dimensional data with historical repair experience and realizes dynamic adjustment through data fusion analysis and machine learning algorithms. In the real-time data-driven optimization link, the system uses a rule engine to analyze parameters such as the preprocessed environmental temperature and humidity gradient, the structural displacement change rate, and the equipment current fluctuation curve. For example, for the temperature regulation strategy of the air conditioning system, by calculating the standard deviation of the temperature distribution in each area of the current floor, the air supply temperature compensation parameter is dynamically generated (compensation value = 0.6×(highest temperature - lowest temperature)), so that the air supply strategy adapts to the real-time environmental differences. In the historical case matching link, the system uses the K-Nearest Neighbor (K-NN) algorithm to retrieve the top 5 cases with the highest similarity to the current abnormal scenario from the historical repair logs (including fields such as fault scenario description, strategy execution steps, repair time consumption, and energy consumption changes) stored in the time series database, and extracts the execution process with the highest average repair efficiency. For example, when dealing with elevator door failures, the system reuses the "three-level troubleshooting process" with the shortest time consumption in historical cases (first detect the door lock sensor signal → then check the control circuit voltage → finally test the wear degree of mechanical components), and adjusts the priority of each troubleshooting step according to the service life and operating load of the current equipment (for example, old equipment is given priority to check mechanical components). The optimized repair plan includes specific control instructions (such as adjusting the fresh air valve opening of the 3rd floor to 85% and maintaining it for 30 minutes), responsible departments and personnel, time window requirements (such as completing the preliminary structural stability assessment within 20 minutes), and safety verification rules (such as confirming that the evacuation status of the area personnel has been completed before execution), and is pushed to the building automation system or the operation and maintenance management platform in real time through the application program interface to realize the intelligent generation and accurate execution of the repair plan.

[0086] The above technical solution is based on the spatial linkage positioning technology of BIM and sensor coordinates, which concretizes the repair target from an abstract abnormal type to a specific physical entity (such as a certain device or structural node on a certain floor), improving the response efficiency of operation and maintenance personnel; in addition, through the real-time data-driven rule engine and the machine learning algorithm of historical case reuse, it solves the technical problem that fixed preset strategies are difficult to adapt to complex scenario changes; moreover, the deep integration of the repair strategy database with the BIM model, equipment ledger, and historical logs forms a full-life cycle closed-loop management system of "monitoring, early warning, repair, and recording", providing data support for building preventive maintenance. Embodiment Seven

[0087] In this embodiment, the intelligent building health monitoring method further includes the following steps: S60: After implementing the repair plan for the intelligent building, monitor the repair effect in real time. By comparing the health status analysis results before and after the repair, evaluate whether the repair effect meets the predetermined standard; if the repair effect does not meet the predetermined standard, adjust the repair plan and re-implement it until the predetermined standard is reached.

[0088] Specifically, step S60 includes: S601: After implementing the repair plan, collect and monitor the multi-dimensional data of the intelligent building in real time.

[0089] After the repair plan of the intelligent building starts to be implemented, the system immediately initiates real-time data collection and monitoring. To ensure a comprehensive and timely reflection of the post-repair situation, the scope of data collection covers the multi-dimensional data of the intelligent building, including environmental data (such as temperature, humidity, air quality, etc.), structural data (such as stress, displacement, structural deformation, cracks, etc. of the building structure), equipment status data (such as operating parameters, energy consumption, etc. of various equipment), and personnel activity data (such as personnel distribution, movement, etc.).

[0090] During the data collection process, the same sensor network as in the previous monitoring is used, and these sensors are reasonably deployed at various key positions in the building. For example, environmental sensors are distributed on different floors and areas to obtain accurate environmental parameters; structural sensors are installed on the main load-bearing structures of the building to monitor the mechanical properties of the structure in real time; equipment sensors are connected to various equipment to collect the operating status information of the equipment; and personnel activity monitoring is achieved through devices such as cameras and access control systems.

[0091] To ensure the real-time and accuracy of the data, the system adopts a high-speed data transmission protocol to transmit the data collected by the sensors to the data processing center in a timely manner. And, to handle the data loss or errors that occur, a data verification and retransmission mechanism is also set up.

[0092] S602: According to the multi-dimensional data, use the health monitoring model to generate the analysis results of the post-repair health status.

[0093] After collecting the post-repair multi-dimensional data, the system inputs these data into the health monitoring model. The health monitoring model will perform a series of processing and analysis on the input multi-dimensional data. First, preprocess the data, including operations such as denoising and normalization, to improve the quality and usability of the data. Then, through the data fusion layer in the model, perform spatio-temporal alignment and dimensional normalization processing on different types of data to generate a standardized data set. Next, the feature extraction layer extracts environmental features, structural features, equipment features, and personnel activity features based on the standardized data set through a convolutional neural network or a graph neural network. Finally, the health assessment layer outputs the analysis results of the post-repair health status such as the environmental condition suitability score, building structure stability index, equipment health percentage, and personnel activity risk level according to these features.

[0094] S603: Compare the analysis results of the pre-repair health status with the analysis results of the post-repair health status to determine whether the repair effect meets the predetermined standards.

[0095] After obtaining the repaired health status analysis result, the system will make a detailed comparison between it and the health status analysis result before repair. To more accurately evaluate the repair effect, the system has preset a series of predetermined criteria, which are formulated according to the design requirements, safety specifications and actual usage needs of the intelligent building.

[0096] For the environmental condition suitability score, if the score before repair is low, the predetermined criterion is set to increase the score by a certain value, such as more than 10 points. For the building structure stability index, the predetermined criterion stipulates that the fluctuation range of the index should be within a small interval, such as ±5%. For the equipment health percentage, the predetermined criterion requires reaching a certain threshold, such as more than 60%. For the personnel activity risk level, the predetermined criterion is set to decrease the level by at least one level.

[0097] The system will conduct a comparative analysis on each evaluation index one by one, calculating the difference or change rate before and after repair. If all evaluation indexes meet the predetermined criteria, it is determined that the repair effect meets the requirements; otherwise, it is determined that the repair effect does not meet the predetermined criteria.

[0098] S604: If the repair effect does not meet the predetermined criteria, adjust the repair plan.

[0099] When it is determined that the repair effect does not meet the predetermined criteria, the system will immediately activate the adjustment mechanism of the repair plan. The basis for adjusting the plan mainly comes from two aspects: one is the historical repair record database, and the other is the expert rule engine.

[0100] The system will retrieve cases similar to the current abnormal scenario from the historical repair record database, adopt the Case-Based Reasoning (CBR) algorithm, analyze the effective repair strategies and adjustment methods in these cases, extract the difference parameters from the current situation, and make a preliminary correction to the existing repair plan.

[0101] The expert rule engine will conduct an in-depth causal analysis on the unqualified health indicators. For example, if the environmental condition suitability score does not meet the predetermined criteria, the expert rule engine will trace the deviation of parameters such as the opening degree of the ventilation system regulating valve and the fresh air volume from the set values of the repair plan, find out the factors affecting the repair effect, and generate corresponding parameter adjustment instructions.

[0102] Based on the analysis results of the above two aspects, the system will comprehensively adjust the repair plan, including adjusting the operating parameters of the equipment, increasing or decreasing certain repair measures, extending the repair time, etc.

[0103] S605: Repair again according to the adjusted repair plan, and repeat steps S601 to S604 until the repair effect meets the predetermined criteria.

[0104] According to the adjusted repair plan, the system will repair the intelligent building again. During the repair process, the system will strictly execute various repair measures according to the adjusted plan to ensure the accuracy and effectiveness of the repair work.

[0105] After the repair is completed, the system will repeat steps S601 to S604 again, that is, collect and monitor the multi-dimensional data of the intelligent building in real time, use the health monitoring model to generate the analysis result of the health status after repair, compare it with the result before repair, and judge whether the repair effect meets the predetermined standard. If it still does not meet the predetermined standard, the system will continue to adjust the repair plan and repair again. This cycle continues until the analysis results of the health status in multiple consecutive monitoring periods (for example, 3 monitoring periods, each period is 1 hour) all meet the predetermined standard.

[0106] In the above intelligent building health monitoring method, the technical solution for repair effect evaluation and plan adjustment constructs a complete and closed-loop repair effect evaluation and adjustment system to ensure that the repair work can achieve the expected goal. After the repair plan is implemented, by collecting multi-dimensional data in real time and using the health monitoring model to generate analysis results, it can comprehensively and accurately reflect the health status of the repaired building. Comparing the results before and after repair and judging the repair effect according to the predetermined standard make the evaluation process objective and scientific. If the repair effect does not meet the standard, the repair plan is adjusted by integrating historical repair records and the expert rule engine, ensuring the rationality and effectiveness of the adjustment. By continuously repeating the process of monitoring, evaluation and adjustment until the repair effect meets the predetermined standard, the quality and efficiency of the repair work are improved, and the potential safety hazards caused by incomplete repair are reduced. Embodiment VIII

[0107] As Figure 3 shown, this embodiment provides an intelligent building health monitoring system, which includes: A data acquisition module, which includes a plurality of sensors arranged at a plurality of target monitoring positions in the intelligent building, configured to obtain multi-dimensional data inside and outside the intelligent building in real time through the plurality of sensors, and the multi-dimensional data includes environmental data, structural data of the intelligent building, equipment status data, and personnel activity data; A data preprocessing module, configured to preprocess the multi-dimensional data to obtain preprocessed multi-dimensional data; A health analysis module, configured to output the health status analysis result of the intelligent building according to the preprocessed multi-dimensional data by using a pre-trained health monitoring model, and the health status analysis result includes the suitability of environmental conditions, the stability of the building structure, the health of equipment, and the risk degree of personnel activities; An early warning module, configured to issue an early warning signal when the health status analysis result indicates that there is a health anomaly in the intelligent building; A repair plan generation module, configured to determine a repair plan for the intelligent building according to the type of the health anomaly, and the repair plan is at least associated with the repair and reinforcement of the intelligent building.

[0108] In a further embodiment, the data acquisition module includes: An environmental data acquisition unit, configured to collect the environmental data inside the intelligent building in real time through environmental sensors, and the environmental data includes temperature, humidity, air quality and light intensity; A structural data acquisition unit, configured to collect the structural data of the intelligent building in real time through structural health monitoring sensors, and the structural data includes displacement, deformation, stress, vibration and crack width of the intelligent building; A device status acquisition unit, configured to collect the device status data of the intelligent building in real time through device status monitoring sensors, and the device status data includes the operation status data of the power system, heating system, ventilation system and air conditioning system; A personnel activity acquisition unit, configured to collect the personnel activity data inside the intelligent building in real time through a personnel positioning and monitoring device, and the personnel activity data includes the distribution density of personnel, the activity trajectory of personnel, and the residence duration of personnel in different areas.

[0109] In a further embodiment, the data preprocessing module includes: A denoising processing unit, configured to perform denoising processing on the multi-dimensional data to obtain purified data; An anomaly detection unit, configured to perform anomaly detection on the purified data to remove anomaly data points and obtain cleaned data; A normalization processing unit, configured to perform normalization processing on the cleaned data to obtain normalized data; A time series analysis unit, configured to perform time series analysis on the normalized data to obtain preprocessed data.

[0110] In a further embodiment, the health analysis module includes: A health monitoring model, including a data fusion layer, a feature extraction layer and a health assessment layer; The data fusion layer is configured to perform spatio-temporal alignment and dimension normalization processing on the preprocessed multi-dimensional data to generate a standardized data set; The feature extraction layer is configured to extract environmental features, structural features, device features and personnel activity features based on the standardized data set through a convolutional neural network or a graph neural network; A health assessment layer, configured to respectively output an environmental condition suitability score, a building structure stability index, a device health percentage, and a personnel activity risk level according to the environmental characteristics, the structural characteristics, the device characteristics, and the personnel activity characteristics.

[0111] As Figure 4 shown, in a further embodiment, the warning module includes: An abnormal type determination unit, configured to determine the health abnormal type of the intelligent building according to the health condition analysis result, where the health abnormal type includes unsuitable environmental conditions, abnormal building structure, equipment failure, and abnormal personnel activities; A warning trigger condition setting unit, configured to set corresponding warning trigger conditions for each health abnormal type, where the warning trigger conditions include that the health condition exceeds a preset threshold; A warning signal sending unit, configured to send a warning signal according to the warning trigger condition.

[0112] In a further embodiment, the repair plan generation module includes: An abnormal target identification unit, configured to identify the abnormal target object corresponding to the health abnormality according to the type of the health abnormality, where the abnormal target object includes the ventilation system, the load-bearing structure, the elevator equipment, or the crowded area of the intelligent building; A repair strategy retrieval unit, configured to retrieve at least one basic repair strategy that matches the type of the health abnormality and the abnormal target object from a preset repair strategy database, where the repair strategy database stores basic repair strategies for different health abnormal types and the abnormal target object; A strategy optimization unit, configured to optimize the retrieved basic repair strategy according to the preprocessed multi-dimensional data and historical repair records, and generate a repair plan for the intelligent building.

[0113] As Figure 3 shown, in a further embodiment, the system further includes: A repair effect evaluation module, configured to, after implementing the repair plan of the intelligent building, monitor the repair effect in real time, and evaluate whether the repair effect meets the predetermined standard by comparing the health condition analysis results before and after the repair; if the repair effect does not meet the predetermined standard, adjust the repair plan and re-implement it until the predetermined standard is reached. When re-implementing, send a re-monitoring instruction to the data collection module.

[0114] In a further embodiment, the repair effect evaluation module includes: A post-repair data collection unit, configured to collect and monitor the multi-dimensional data of the intelligent building in real time after implementing the repair plan; A post - repair analysis unit, configured to generate a post - repair health status analysis result according to the multi - dimensional data using a health monitoring model; An effect comparison unit, configured to compare the pre - repair health status analysis result with the post - repair health status analysis result to determine whether the repair effect meets a predetermined standard; A solution adjustment unit, configured to adjust the repair solution if the repair effect does not meet the predetermined standard; A loop execution unit, configured to perform repair again according to the adjusted repair solution and repeat the steps of data collection, analysis, comparison, and adjustment until the repair effect meets the predetermined standard.

[0115] In a further embodiment, the system further includes: An early - warning linkage execution module, configured to perform corresponding operations according to the early - warning type after the early - warning signal is sent: When an environmental - type early - warning signal is triggered, link to turn on the ventilation and purification equipment and adjust the temperature and humidity control system; When a structure - type early - warning signal is triggered, automatically lock the entrances and exits of the abnormal area and prohibit heavy equipment from running; When an equipment - type early - warning signal is triggered, switch to the standby equipment and generate a maintenance work order to be pushed to the maintenance unit; When a personnel - type early - warning signal is triggered, start the emergency broadcast system and turn on the emergency evacuation indicator lights.

[0116] An embodiment of the present invention also provides a computer - readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements any one of the above - mentioned methods.

[0117] The present invention also provides an electronic device. The electronic device according to the embodiment of the present invention includes: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method provided by the present invention.

[0118] Next, refer to Figure 5 , which shows a schematic structural diagram of a computer system 800 of an electronic device suitable for implementing the embodiment of the present invention. Figure 5 The shown electronic device is only an example and should not bring any limitation to the functions and usage scope of the embodiment of the present invention.

[0119] As Figure 5As shown, computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 502 or programs loaded from a storage section 505 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the computer system 500 are also stored. The CPU 501, ROM 502, and RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0120] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as required. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as required so that a computer program read therefrom is installed into the storage section 505 as required.

[0121] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A health monitoring method for an intelligent building, characterized in that, The method includes the following steps: S10: Real-time obtain multi-dimensional data inside and outside the intelligent building through multiple sensors set at multiple target monitoring positions of the intelligent building. The multi-dimensional data includes environmental data, structural data of the intelligent building, equipment status data, and personnel activity data. The structural data includes displacement, deformation, stress, vibration, and crack width of the intelligent building; S20: Preprocess the multi-dimensional data to obtain preprocessed multi-dimensional data; S30: According to the preprocessed multi-dimensional data, use a pre-trained health monitoring model to output an analysis result of the health status of the intelligent building. The health status analysis result includes the suitability of the environmental conditions, the stability of the building structure, the health of the equipment, and the risk level of personnel activities; S40: When the health status analysis result indicates that there is a health anomaly in the intelligent building, issue a warning signal; S50: Determine a repair plan for the intelligent building according to the type of the health anomaly. The repair plan is at least associated with the repair and reinforcement of the building structure of the intelligent building.

2. The health monitoring method of the intelligent building according to claim 1, characterized in that, Step S10 includes: S101: Real-time collect the environmental data inside the intelligent building through environmental sensors. The environmental data includes temperature, humidity, air quality, and light intensity; S102: Real-time collect the structural data of the intelligent building through structural health monitoring sensors; S103: Real-time collect the equipment status data of the intelligent building through equipment status monitoring sensors. The equipment status data includes the operating status data of the power system, heating system, ventilation system, and air conditioning system; S104: Real-time collect the personnel activity data inside the intelligent building through a personnel positioning and monitoring device. The personnel activity data includes the distribution density of personnel, the activity trajectory of personnel, and the residence duration of personnel in different areas.

3. The health monitoring method of the intelligent building according to claim 1, wherein the environmental data is collected through temperature and humidity sensors, carbon dioxide concentration sensors, light intensity sensors, and air quality sensors distributed in each functional area of the intelligent building; the structural data of the intelligent building is collected through pressure sensors, strain sensors, vibration sensors, and displacement sensors set at the target structural parts of the intelligent building. The target structural parts include beams, columns, walls, foundations, and roofs; the equipment status data is collected through current sensors, voltage sensors, temperature sensors, and vibration sensors installed on air conditioning systems, elevators, lighting equipment, water supply and power supply equipment, and ventilation equipment; the personnel activity data is collected through cameras, infrared sensors, and access control systems deployed in public areas, entrances and exits, and stairwells of the intelligent building.

4. The health monitoring method of the intelligent building according to claim 1, characterized in that, Step S30 includes: S301: Input the preprocessed multi-dimensional data into the health monitoring model. The health monitoring model includes a data fusion layer, a feature extraction layer, and a health assessment layer; S302: The data fusion layer performs spatio-temporal alignment and dimensional normalization processing on the multi-dimensional data to generate a standardized data set; S303: The feature extraction layer extracts environmental features, structural features, equipment features, and personnel activity features based on the standardized data set through a convolutional neural network or a graph neural network; S304: The health assessment layer outputs an environmental condition suitability score, a building structure stability index, an equipment health percentage, and a human activity risk level, respectively, based on the environmental characteristics, the structural characteristics, the equipment characteristics, and the human activity characteristics.

5. The health monitoring method of the intelligent building according to claim 1, characterized in that, Step S50 includes: when the type of health anomaly is a structural anomaly of the intelligent building, matching a corresponding repair solution according to the type of the structural anomaly of the intelligent building: When the type of structural abnormality is structural deformation, a first repair solution is performed based on the deformation amount; When the type of structural anomaly is a structural crack, a second repair solution is performed based on the crack width classification; When the type of structural anomaly is structural material performance degradation, executing a third repair solution based on the material type; Among them, the detection of structural deformation adopts a total station, a level, a laser scanner, a strain sensor and a displacement sensor; the detection of structural cracks adopts a crack width gauge and an ultrasonic flaw detector; in the detection of material performance degradation, a rebound hammer or a carbonation depth gauge is used for concrete, and a magnetic particle flaw detector or an ultrasonic thickness gauge is used for steel structure.

6. The health monitoring method of the intelligent building according to claim 5, characterized in that, The first repair solution includes: For mild deformation within the first deformation range, the following operations were performed: adjusting the load configuration, re-tightening the connectors, and attaching carbon fiber cloth to the surface of the deformed area; For moderate deformation in the second deformation range, the following operations are performed: set up steel supports or jacks to transfer the load, implant steel bars into the concrete structure and pour concrete, and weld stiffeners to the steel structure for reinforcement; For severe deformation in the third deformation range, the following actions are performed: initiate personnel evacuation and warning measures, use external prestressed tendons to apply reverse force, and remove irreparable structural components; The first deformation range, the second deformation range, and the third deformation range increase in sequence.

7. The health monitoring method of the intelligent building according to claim 5, characterized in that, The second repair solution includes: For cracks in the first width range, epoxy putty is used for surface sealing treatment; For cracks in the second width range, pressure grouting is used to fill with epoxy resin and adhere steel plates or carbon fiber cloth; For cracks in the third width range, foundation reinforcement treatment or a repair method combining grouting and steel plate bonding is used; The first width range, the second width range, and the third width range increase in sequence.

8. The health monitoring method of the intelligent building according to claim 5, characterized in that, The third repair solution includes: To address concrete deterioration, perform the following: apply concrete protective agent to delay carbonization, remove rust and paint after chiseling out rusted areas, and use polymer concrete to repair chiseled areas; For deterioration of steel structures, perform the following operations: sandblasting to remove rust, then apply anti-corrosion paint, welding steel plates or pasting carbon fiber cloth.

9. The health monitoring method of the intelligent building according to claim 1, wherein The method further comprises the following steps: S60: After implementing the repair plan for the intelligent building, the repair effect is monitored in real time. By comparing the health condition analysis results before and after the repair, it is evaluated whether the repair effect meets the predetermined standard. If the repair effect does not meet the predetermined standard, the repair plan is adjusted and re-implemented until the predetermined standard is reached.

10. A health monitoring system for an intelligent building, characterized in that, The system includes: A data acquisition module, which includes a plurality of sensors arranged at multiple target monitoring positions in the intelligent building, configured to obtain multi-dimensional data inside and outside the intelligent building in real time. The multi-dimensional data includes environmental data, structural data of the intelligent building, equipment status data, and personnel activity data. The structural data includes the displacement, deformation, stress, vibration, and crack width of the intelligent building. A data preprocessing module, configured to preprocess the multi-dimensional data to obtain preprocessed multi-dimensional data. A health analysis module, configured to output the health condition analysis result of the intelligent building according to the preprocessed multi-dimensional data by using a pre-trained health monitoring model. The health condition analysis result includes the suitability of the environmental conditions, the stability of the building structure, the health of the equipment, and the risk level of personnel activities. An early warning module, configured to issue an early warning signal when the health condition analysis result indicates that there is a health abnormality in the intelligent building. A repair plan generation module, configured to determine the repair plan for the intelligent building according to the type of the health abnormality. The repair plan is at least associated with the repair and reinforcement of the building structure of the intelligent building.

Images