A safety monitoring system and method for a building curtain wall

Abstract

The application relates to the technical field of safety monitoring, in particular to a safety monitoring system and a monitoring method for a building curtain wall, which comprises a non-invasive sensor array composed of multiple independent monitoring nodes, each monitoring node comprising a sensing module for collecting curtain wall physical state parameters and a connecting interface for realizing physical connection with the curtain wall surface, the connecting interface adopts a reversible connection mode based on adsorption force or magnetic force, so that the whole monitoring node can be installed and removed without damaging the curtain wall structure; a data aggregation and communication gateway for receiving monitoring data from the non-invasive sensor array and performing preliminary packaging and protocol conversion on the data; and a cloud or local data analysis platform for receiving data from the data aggregation and communication gateway, executing a safety state analysis algorithm, and generating visual results and early warning information. The application improves the deployment universality, maintainability and historical building applicability of the system.

Description

Technical Field

[0001] This application relates to the technical field of safety monitoring, and in particular to a safety monitoring system and method for building curtain walls. Background Technology

[0002] As the external envelope of modern buildings, the safety of building curtain walls is directly related to the overall stability of the building and the safety of life and property of users. Therefore, real-time monitoring of building curtain walls to detect potential safety hazards at an early stage has become an important technological direction in the field of building maintenance. Monitoring systems typically use sensors to collect status data of the curtain wall, such as deformation, vibration, or temperature changes, thereby assessing its safety condition.

[0003] In related technologies, a common monitoring method is to install physical sensors on the surface or structure of the curtain wall, such as fixing strain gauges by drilling holes or attaching accelerometers using adhesives. These sensors are directly attached to the curtain wall and transmit the collected data to a processing unit for analysis via wired or wireless means to achieve continuous monitoring of the curtain wall's safety.

[0004] However, this installation method has significant technical drawbacks. Because it requires drilling or pasting into the curtain wall, the installation process is invasive and can damage the structural integrity and aesthetics of the curtain wall. Especially for historic buildings or existing projects, this destructive installation may introduce new stress points or affect sealing performance, limiting the application scope and deployment flexibility of the monitoring system. Summary of the Invention

[0005] To address at least one of the aforementioned technical problems, this application provides a safety monitoring system and method for building curtain walls.

[0006] In a first aspect, this application provides a safety monitoring system for building curtain walls, employing the following technical solution, including: The non-invasive sensor array consists of multiple independent monitoring nodes. Each monitoring node includes a sensing module for collecting physical state parameters of the curtain wall and a connection interface for physically connecting with the curtain wall surface. The connection interface adopts a reversible connection method based on adsorption or magnetism, which allows the entire monitoring node to be installed and removed without damaging the curtain wall structure. A data aggregation and communication gateway is used to receive monitoring data from the non-invasive sensor array and to perform preliminary data packaging and protocol conversion. And a cloud or local data analysis platform, used to receive data from the data aggregation and communication gateway, execute security status analysis algorithms, and generate visualization results and early warning information; Each monitoring node in the non-invasive sensor array has an independent identification and is bound to a specific physical location on the curtain wall surface.

[0007] By adopting the above technical solutions, this system constructs a non-invasive sensor array using reversible connection interfaces based on adsorption or magnetism, completely eliminating traditional destructive installation methods such as drilling and bonding. While achieving full-surface, location-based monitoring, it absolutely guarantees the integrity and aesthetics of the curtain wall structure. The system architecture clearly defines the three levels of sensing, aggregation, and analysis, and introduces a node identity and location binding mechanism, providing a hardware foundation for high-precision, traceable monitoring and significantly improving the system's deployment versatility, maintainability, and applicability to historical buildings.

[0008] In one possible implementation, each monitoring node in the non-invasive sensor array further includes a micro-energy management unit that integrates an environmental energy harvesting device and a rechargeable energy storage element; wherein the environmental energy harvesting device is configured to harvest energy from light, temperature difference or vibration in the environment where the monitoring node is located, and to power the sensing module and communication components, thereby enabling the long-term self-sustaining operation of the monitoring node.

[0009] By adopting the above technical solution and integrating a micro-energy management unit into each non-intrusive node, especially by combining various environmental energy harvesting technologies, the problem of continuous power supply for passive deployment nodes is fundamentally solved. This frees the monitoring nodes from dependence on fixed power sources or frequent battery replacements, enabling them to operate immediately upon deployment. This greatly expands the system's application capabilities in scenarios without external power sources and further reduces the complexity and cost of operation and maintenance throughout the entire lifecycle.

[0010] In one possible implementation, the connection interface specifically includes a flexible vacuum suction cup array or a programmable electromagnetic adsorption module; wherein, the flexible vacuum suction cup array can adaptively conform to the curved or uneven surface of the curtain wall to form a sealed negative pressure zone, and the programmable electromagnetic adsorption module can generate a strong adsorption force when powered on and completely release when powered off.

[0011] By adopting the above technical solutions, two specific and efficient implementation methods for reversible connection interfaces are further defined: flexible vacuum suction cups and controllable electromagnets. The flexible vacuum suction cups solve the sealing problem of curved surface installation through adaptive fitting, while the controllable electromagnets provide stable and easily remotely controlled adsorption force. Both of these specific structures refine the non-invasive reversible connection, achieving high-strength, repeatedly detachable installation effects in a reliable physical manner, which are key technical guarantees for ensuring system stability and practicality.

[0012] Secondly, this application provides a safety monitoring method for building curtain walls, applicable to the system described in the first aspect and any possible implementation thereof, the method comprising: Multiple monitoring nodes are reversibly attached to the surface of the curtain wall to collect state parameter data of the curtain wall at different locations in parallel. The status parameter data collected by each monitoring node, along with its identification information, are aggregated to the data aggregation point via a wireless communication network. Based on a pre-set safety assessment model, the aggregated status parameter data with location markers are analyzed and calculated to determine the overall and local safety status of the curtain wall. Based on the assessment of the safety status, monitoring reports and early warning prompts are dynamically generated and output for different locations or different risk levels.

[0013] By adopting the above technical solution, the core of this method lies in its integration with a non-destructive hardware system, enabling a complete process of location-based data acquisition, aggregation, and location correlation analysis. By emphasizing location correlation analysis based on node identification, the monitoring results not only reflect whether a problem exists but also precisely indicate where the problem is located, greatly improving the accuracy of monitoring and the guidance of maintenance actions. The entire method completely avoids physical modification to the building itself, representing a complete operational embodiment of the reversible, non-intrusive monitoring concept.

[0014] In one possible implementation, the parallel acquisition step further includes: When collecting the status parameter data, each monitoring node simultaneously collects the interference parameter data of its local environment.

[0015] In one possible implementation, the analysis and calculation steps specifically include: The interference parameter data is used to compensate and correct the corresponding state parameter data in real time, so as to eliminate the error caused by environmental fluctuations to the monitoring data. Perform a safety status assessment based on the corrected status parameter data.

[0016] By adopting the above technical solution, this method adds steps for synchronous environmental perception and real-time data compensation, improving the inherent reliability of monitoring in complex outdoor environments such as drastic temperature changes and wind and rain interference. By performing paired acquisition and preliminary fusion of status data and interference data locally at each node, environmental noise is effectively suppressed, allowing subsequent safety analysis to be based on a cleaner and more reliable signal foundation. This significantly reduces the probability of false alarms and missed alarms, enhancing the credibility of monitoring conclusions.

[0017] In one possible implementation, the step of summarizing to the data aggregation point specifically includes: A multi-hop ad hoc network communication protocol is adopted, in which some monitoring nodes are configured as routing nodes, which can forward data from other adjacent monitoring nodes, thereby constructing a mesh data transmission network with redundant paths covering the entire curtain wall area within the wireless communication network.

[0018] By adopting the above technical solution and introducing a multi-hop self-organizing network communication mechanism, the monitoring node network is endowed with self-organizing and self-healing communication capabilities. This effectively overcomes the problems of limited transmission distance and easy signal blockage at single points, ensuring the robustness and integrity of monitoring data transmission in complex building facade environments. Furthermore, the mesh redundancy path enhances network reliability; a single node communication failure will not lead to data link interruption, thus achieving stable and reliable aggregation of large-scale, high-density sensor array data.

[0019] In one possible implementation, the method further includes: Regularly or trigger-based assessments of the data quality, signal strength, and energy levels of each monitoring node; When a node's performance metric is found to be below a preset threshold, the node is automatically marked as requiring maintenance in the visualization results, and a maintenance work order is generated.

[0020] By adopting the above technical solutions, the monitoring target is extended from the curtain wall to the monitoring network itself, realizing the system's self-sensing and self-maintenance early warning. By automatically monitoring the health status of nodes and generating work orders based on their physical locations, passive maintenance is transformed into proactive preventive maintenance, greatly improving the online rate and data availability of the entire monitoring system, reducing the risk of monitoring blind spots caused by equipment failure, and enhancing the system's long-term stability and management efficiency.

[0021] In one possible implementation, the analysis and computation steps are executed using a distributed edge collaborative computing architecture, specifically including: The first-level analysis is performed at the data aggregation and communication gateway, which includes data filtering, feature extraction, and rapid anomaly detection. Only when an anomaly is detected or at a fixed period, the processed feature data or anomaly events are uploaded to the cloud or local data analysis platform for second-level in-depth analysis and trend prediction.

[0022] By adopting the above technical solution and implementing distributed edge collaborative computing, the data processing flow has been optimized. Real-time anomaly detection, which requires high precision, is decentralized to the network edge gateway, significantly reducing reliance on cloud bandwidth and data analysis latency, achieving near real-time alerts. The cloud, on the other hand, focuses on in-depth analysis requiring large datasets and complex models. This division of labor and collaboration not only alleviates the computational burden on the central platform but also improves the overall system response speed and scalability.

[0023] Thirdly, this application provides an electronic device including a memory and a processor, wherein the memory is used to store computer program code, and the processor is used to execute the computer program code stored in the memory to implement the methods in the first aspect and any one of the first aspects, or in the second aspect and any possible implementation of the second aspect.

[0024] Fourthly, this application provides a computer-readable storage medium storing a computer program or instructions that, when executed, implement the methods described in the first aspect and any one thereof, or the second aspect and any possible implementation thereof.

[0025] In summary, this application includes at least one of the following beneficial technical effects: 1. It achieves absolutely non-destructive reversible installation monitoring. By adopting a reversible connection interface based on adsorption or magnetism, monitoring nodes can be firmly attached to any surface of the curtain wall without drilling or bonding, and can be removed without damage as needed. This fundamentally solves the industry problem of traditional monitoring technologies damaging the integrity and aesthetics of building structures, enabling the technology to be applied to historical buildings, already decorated curtain walls, and other fields.

[0026] 2. A self-sufficient, long-term autonomous monitoring node was constructed, whereby each monitoring node integrates a micro-energy management unit that harvests energy from the environment, enabling the node to operate under its own power. This design completely eliminates the dependence on external power sources or frequent battery replacements, solving the most critical issue of sustained power supply in large-scale wireless sensor network deployments and significantly reducing the overall lifecycle maintenance costs.

[0027] 3. A collaborative intelligent architecture has been formed, integrating real-time edge early warning and in-depth cloud analysis. This architecture utilizes distributed edge computing to rapidly process data and determine anomalies in real time at the network edge, close to sensors, achieving near-local alarm processing. Simultaneously, feature data and complex model analysis are handled by the cloud platform. This architecture differentiates between the real-time and complex aspects of processing tasks, ensuring immediate response to security risks while reducing the load on the central system. This provides crucial support for the feasibility and efficiency of ultra-large-scale, high-density monitoring networks. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating a safety monitoring method for building curtain walls provided in an embodiment of this application.

[0029] Figure 2 This is a structural schematic diagram of a safety monitoring system for a building curtain wall provided in an embodiment of this application.

[0030] Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0031] The technical solutions in this application will now be described with reference to all the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0032] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. "And / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Furthermore, in the description of the embodiments of this application, "plural" or "multiple" refers to two or more than two.

[0033] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this embodiment, unless otherwise stated, "a plurality of" means two or more.

[0034] The terminology used in the following embodiments is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to also include expressions such as “one or more,” unless the context clearly indicates otherwise. It should also be understood that in the following embodiments of this application, “at least one” and “one or more” refer to one, two, or more than two.

[0035] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "one embodiment," "some embodiments," "another embodiment," "other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0036] This application provides a method for safety monitoring of building curtain walls, executed by an electronic device. This electronic device can be a standalone physical electronic device, a cluster of multiple physical electronic devices, a distributed system, or a cloud electronic device providing cloud computing services. This application does not impose limitations on this method. Figure 1 As shown, the method includes: S1. Multiple monitoring nodes are reversibly attached to the surface of the curtain wall to collect state parameter data of the curtain wall at different locations in parallel.

[0037] Specifically, the reversible attachment method is implemented by integrating a connection interface module at the bottom of the shell of each monitoring node.

[0038] This module comprises at least two physical forms. The first is a flexible vacuum suction cup array, consisting of multiple miniature, highly elastic silicone suction cups. Each suction cup has a miniature cavity inside and is connected to a miniature vacuum pump via a capillary tube. When the node is attached to the curtain wall surface, the silicone suction cup deforms due to its elasticity to adapt to the slight unevenness or curvature of the curtain wall. The miniature vacuum pump is activated to extract air from the cavity, creating negative pressure and thus generating suction force.

[0039] The second type is the electromagnetic adsorption module. Its core consists of a planar electromagnet array encapsulated within the node's outer shell and a magnetically conductive sheet that can be attached to the corresponding position on the curtain wall. This sheet is extremely thin and can be temporarily pasted in advance with non-destructive adhesive or embedded within the curtain wall glass interlayer. It generates or releases a powerful magnetic force by controlling the energization and de-energization of the electromagnets through the internal circuitry of the node.

[0040] In addition, the node integrates a miniature inertial measurement unit for collecting vibration and acceleration data, a miniature strain sensing unit that indirectly reflects the curtain wall strain by sensing the micro-deformation of the node's shell, a temperature sensor, and a wireless communication module. Each node is assigned a globally unique digital identification code at the factory and stored in its built-in microcontroller storage unit.

[0041] Furthermore, to achieve parallel data acquisition, after the system is powered on, all monitoring nodes successfully attached to the curtain wall surface independently and simultaneously start their sensing units to acquire data according to preset, staggered time slots or frequency bands.

[0042] The acquisition process is scheduled by the microcontroller within the node, either at a fixed cycle such as once per minute or triggered by events such as vibration exceeding a threshold.

[0043] The key technical feature is that each time raw sensor data is collected, such as triaxial acceleration values ​​and temperature values, it is forcibly bound to the node's unique identification code during the initial packaging process within the node, forming a basic data packet with identity, timestamp, and sensor data structure. This binding operation is completed locally on the node, ensuring data traceability from the source.

[0044] Based on this, the design can physically divide the huge curtain wall monitoring surface into multiple independent, fully functional sub-units / nodes, and pre-embed location information in the initial stage of data generation.

[0045] In summary, firstly, the two connection methods, flexible vacuum suction cup and electromagnetic adsorption, completely avoid permanent fixing methods such as drilling, nailing, and chemical bonding, and truly achieve physical reversibility and non-invasiveness.

[0046] For example, installers can quickly deploy nodes as if they were placing a piece of equipment, and remove them at any time without damage, which is crucial for protecting the building itself, adapting to historical buildings, or meeting temporary monitoring needs.

[0047] Secondly, by strongly binding a unique identifier to the node hardware and encapsulating it with sensor data at the source of data generation, the system can clearly identify which physical node the data comes from, regardless of where the data flows through in subsequent systems.

[0048] During system initialization, by manually or automatically associating each node's identifier with its specific installation location on the building facade drawings through handheld terminal scanning or data entry (e.g., "East facade, 3rd floor, vertical keel No. 2"), the identifier in the data packet becomes equivalent to a precise spatial coordinate. This allows subsequent analysis not only to detect anomalies but also to immediately pinpoint specific physical points or areas on the curtain wall, providing guidance for precise maintenance and solving the pain point of traditional monitoring where a problem is known but its exact location is unknown.

[0049] In some embodiments, the parallel acquisition step in S1 further includes: S101. When collecting status parameter data, each monitoring node simultaneously collects interference parameter data of its local environment.

[0050] Specifically, the synchronous acquisition step is implemented by integrating an additional set of environmental sensing units on the hardware circuit board of each monitoring node, in addition to the sensing units used to sense the physical state of the curtain wall itself, specifically for sensing environmental disturbances.

[0051] This unit typically includes at least one high-precision temperature sensor, such as a digital temperature sensing chip with an accuracy of ±0.5°C; a humidity sensor, which may be selected to sense the effect of air humidity on certain sensors or structural adhesives; an atmospheric pressure sensor, which may be selected to assist in judging changes in weather systems; and a triaxial micro accelerometer, which may be the same device in hardware as the accelerometer used to monitor curtain wall vibration, but is configured in software to operate at different frequencies or modes, specifically for sensing low-frequency, large-amplitude node sway caused by strong winds, thus serving as an indirect measure of wind load disturbance.

[0052] The above environmental sensors and status sensors share the same microcontroller and power system, but their data acquisition channels are independent.

[0053] Furthermore, the software and timing logic implementation of the synchronous acquisition step is as follows: in the microcontroller of the monitoring node, the firmware is programmed as a multi-task acquisition scheduler.

[0054] When a data acquisition cycle arrives or an event is triggered, the scheduler does not sequentially acquire state parameters first and then interference parameters. Instead, it issues a synchronous acquisition command. This command, via the internal bus, simultaneously activates the analog-to-digital converters of all sensing units, including state sensing units and environmental sensing units. The analog signals captured by all sensors are synchronously converted into digital signals.

[0055] The microcontroller then packages the digital readings from different physical channels according to a predetermined data frame format. In this data frame, state parameter data and interference parameter data from the same timestamp and the same physical node are encapsulated together, and together with the node's unique identifier and high-precision timestamp, they form a complete data packet with spatiotemporal and environmental context.

[0056] Thus, the characteristics of structural response and environmental factors are combined into the same operation or data object, and the fusion of multi-dimensional information is completed at the source of data generation, preparing the necessary information for subsequent accurate analysis.

[0057] In this embodiment, the method further includes: S2. The status parameter data collected by each monitoring node, along with its identification information, are aggregated to the data aggregation point via a wireless communication network.

[0058] Specifically, the wireless communication network is constructed based on a hybrid network topology, and the wireless communication module built into each monitoring node is configured to switch between two modes.

[0059] In most cases, gateway devices act as end nodes, communicating directly with the nearest data aggregation point. Gateways are typically deployed on building rooftops, windowsills, or interior locations near curtain walls, offering enhanced power and communication capabilities.

[0060] For nodes whose signals are too weak to directly connect to the gateway due to building obstruction, the system activates a second mode: multi-hop self-organizing network mode. In this mode, some monitoring nodes that are pre-configured or dynamically elected as routing nodes based on signal strength, in addition to collecting their own data, are also responsible for forwarding data packets from other leaf nodes within their communication range. All data packets follow a unified network layer protocol, which includes information such as the source node's identity, hop count, and final destination.

[0061] Furthermore, the specific software communication logic of its aggregation process is as follows: after the monitoring node prepares a data packet with an identity identifier, it first adds a network layer header to it in the local communication stack.

[0062] Then, based on its stored neighbor table information and the current channel quality, the node chooses whether to send the packet directly to the gateway or to a relay routing node. If relay is selected, the packet will undergo one or more hops before finally reaching the gateway.

[0063] After receiving the data packet, the gateway will perform protocol parsing, data verification and format standardization, and cache and batch package data packets from different nodes and at different times.

[0064] The gateway then uploads the packaged batch data to the data receiving server in the cloud or local data center via its WAN uplink.

[0065] Throughout the process, the core payload within the data packet—the binding relationship between the node identifier and the sensor data—remains unchanged and is transmitted level by level. Furthermore, the network structure dynamically adjusts according to the environment to optimize communication paths, and data packet acknowledgment and retransmission mechanisms ensure the reliability of data transmission.

[0066] Based on this, the design of the hybrid wireless network effectively solves the problem of full signal coverage in large or complex shaped curtain wall scenarios.

[0067] Among them, the multi-hop self-organizing network capability ensures that even in corners, recesses, or areas with severe signal blockage, the data of the monitoring node can be reliably transmitted through neighboring nodes, eliminating communication blind spots and ensuring the integrity of all monitoring data.

[0068] As a local aggregation center, the gateway performs preliminary cleaning and organization of data, reducing the access pressure on the cloud server.

[0069] Most importantly, the entire transmission chain strictly maintains the identity identifier and data binding relationship. No matter how many hops the data goes through, its source identity remains clear.

[0070] Once the cloud server receives the data, it can instantly convert a string of numerical identifiers into actual locations such as "5th floor glass panel on the north facade" by querying the node identity and physical location mapping database established during system initialization. This provides a solid and reliable data foundation for the subsequent S3 step of analysis and calculation based on precise location, making location correlation analysis not a post-hoc inference, but an inherent characteristic built on the underlying data transmission architecture.

[0071] In some embodiments, to ensure the robustness and integrity of monitoring data transmission in complex building facade environments, and to achieve stable and reliable aggregation of large-scale, high-density sensor array data, the aggregation to the data aggregation point in step S2 specifically includes: S201. A multi-hop self-organizing network communication protocol is adopted, in which some monitoring nodes are configured as routing nodes, which can forward data from other adjacent monitoring nodes, thereby constructing a mesh data transmission network with redundant paths covering the entire curtain wall area within the wireless communication network.

[0072] Specifically, the multi-hop ad hoc network communication protocol relies on selecting a low-power wireless communication chipset that supports mesh networking functions for each monitoring node and developing the corresponding embedded firmware for the nodes.

[0073] The nodes in the network are divided into three logical roles: end nodes, routing nodes, and the coordinator, which is the data aggregation point / gateway. After all nodes are powered on, they first perform a channel scan and attempt to join a network established by the coordinator. The coordinator is responsible for allocating network short addresses and managing network keys.

[0074] In terms of hardware, routing nodes and terminal nodes can be the same in terms of core communication hardware, but the firmware of routing nodes is enabled with full protocol stack routing functions, and its microcontroller needs to have slightly stronger processing power and more RAM in order to maintain routing tables, neighbor tables and perform store-and-forward of data packets.

[0075] The end-node firmware has simplified functions, primarily responsible for its own data collection and communication with the parent node, either the routing node or the coordinator. The coordinator, on the other hand, possesses the strongest processing power and a stable power supply, runs complete network and application layer protocols, and has a WAN uplink interface.

[0076] Furthermore, the software communication logic and self-organizing process for constructing a mesh data transmission network with redundant paths are as follows: After network initialization, the coordinator broadcasts a network beacon. Upon receiving the beacon, the monitoring node selects and associates with a parent-child link establishment process, choosing the parent node with the highest received signal strength index (RSSI) and thus the best signal quality. This parent node can be the coordinator or other routing nodes already in the network, thereby forming the initial tree topology.

[0077] Subsequently, routing nodes periodically broadcast routing announcements, declaring their existence and path costs. Routing nodes and end nodes listen for these announcements and dynamically update and maintain their routing tables accordingly. These tables may record multiple next-hop options and their path costs, such as hop count and link quality, for the target address leading to the coordinator.

[0078] When a monitoring node needs to send a data packet, its network layer protocol stack queries the routing table and selects the optimal path for transmission. Each time a data packet passes through a routing node, that node checks the destination address and hop count limit in the packet header and determines the next hop based on its current routing table.

[0079] If the next node on the preferred path is unresponsive, the routing node can attempt to retransmit via an alternative path or initiate local route repair to find a new neighbor node and establish a link, according to the protocol. Thus, the network topology and paths can dynamically change based on link status; and by maintaining multiple potential paths, it mitigates single points of failure, ensuring the overall reliability of the function.

[0080] In summary, in complex building facade environments, metal frames, thick concrete, and special coatings can severely hinder the straight-line propagation of wireless signals, causing some nodes to be unable to communicate directly with the remote coordinator.

[0081] The multi-hop self-organizing network mechanism decomposes the communication distance problem into multiple short-distance hops, which are relayed by intermediate nodes, effectively penetrating or bypassing physical obstacles and ensuring that data from every node in every corner has a path to reach, thereby guaranteeing the integrity of data collection.

[0082] Redundant paths are crucial in mesh networks. In traditional point-to-point or star networks, if a routing node or a link fails, all data of its downstream nodes will be permanently lost. However, in this implementation, nodes typically know multiple potential parent nodes or next-hop nodes.

[0083] When the main path is interrupted due to node failure, temporary obstruction, or interference, the network protocol can quickly activate backup paths or rebuild local routes to achieve self-healing, ensuring the continuity of data flow, which greatly improves robustness.

[0084] For large-scale, high-density deployments, sending all data directly to a single coordinator can lead to signal congestion and increased collisions around the coordinator. Multi-hop networks distribute the communication load across a large number of routing nodes, spatially distributing and relaying data traffic. The coordinator only needs to handle the final converged traffic, which significantly improves the overall capacity and scalability of the network, making it possible to manage ultra-large-scale sensor arrays.

[0085] Ultimately, the above implementation methods collectively ensure that no matter how complex the curtain wall geometry or how large the number of nodes, the monitoring data can be stably, reliably, and completely collected at the processing center, laying a solid data foundation for subsequent high-precision analysis.

[0086] In this embodiment, the method further includes: S3. Based on the pre-set safety assessment model, analyze and calculate the aggregated status parameter data with location identifiers to determine the overall and local safety status of the curtain wall.

[0087] Specifically, the pre-built security assessment model is a software analysis engine deployed on a cloud server or a high-performance local server, which contains multiple parallel analysis sub-modules.

[0088] The core sub-modules include: a threshold comparison module, which sets static thresholds and adaptive thresholds based on dynamic learning from historical data for parameters such as vibration amplitude, strain change, and temperature gradient; a trend analysis module, which uses time series analysis algorithms such as moving averages and exponential smoothing to identify the changing trends of parameters at specific locations, such as slowly increasing deformation; a correlation analysis module, which analyzes the synchronicity or abnormal correlation of data changes at adjacent or structurally related locations, like multiple nodes on a keel; and a pattern recognition module, which uses a trained lightweight machine learning model to identify whether vibration spectrum features or data combination patterns contain known hazardous patterns, such as specific frequency characteristics of glass breakage or specific vibration signals of loose connectors.

[0089] Furthermore, the specific process of this analysis and calculation step is as follows: the data receiving server unpacks the batch data from the gateway and distributes it to the corresponding data storage area according to the node identity identifier in the data packet, usually indexed by node or region.

[0090] The safety assessment model initiates analysis at predetermined intervals, such as every 5 minutes, or as triggered by the receipt of vibration event data. During analysis, the model first extracts recent historical data of the target node and its adjacent nodes from the database, combining it with real-time data to form a data slice containing spatiotemporal dimensions. Then, each analysis submodule performs calculations on this data slice in parallel.

[0091] For example, the threshold module determines whether the current value exceeds the limit; the trend module determines whether the deformation rate has significantly accelerated in the past hour; and the correlation module determines whether similar abnormal vibrations have occurred at adjacent nodes. The outputs of each module, namely the levels of normal, attention, warning, and danger, and their confidence levels, are fed into a comprehensive decision-making unit.

[0092] Based on preset rules, such as "if three adjacent nodes reach the 'warning' level simultaneously, the entire area is upgraded to 'dangerous'", the integrated decision-maker ultimately generates a safety status judgment for the node's location and higher-level areas.

[0093] Based on this, by deploying a multi-layered, multi-algorithm security assessment model, this step enables the system to go beyond simple over-limit alarms and achieve intelligent diagnosis.

[0094] For example, a single momentary vibration may be caused by a gust of wind and is considered normal, while a continuous micro-vibration in a specific frequency band may be caused by a loose screw and is considered an early warning. The correlation module and the pattern recognition module can effectively distinguish these situations, greatly reducing false alarms.

[0095] More importantly, because the input data naturally carries precise location identifiers, all analysis and calculation results, whether it is threshold exceeding, trend anomaly, or correlation anomaly, can be immediately assigned a precise spatial label.

[0096] This allows the system to output not only a general conclusion that the curtain wall is dangerous, but also a specific and actionable judgment such as "the fastening connection of node 2 in section B on the 3rd floor of the east facade shows a loosening trend, risk level: medium".

[0097] This location-based, refined analysis breaks down safety status assessments from an overall perspective to the component level, enabling deeper monitoring from surface to point. It provides maintenance personnel with precise action guidelines, improving maintenance efficiency and safety.

[0098] In some embodiments, in order to reduce the probability of false alarms and false negatives and enhance the reliability of monitoring conclusions, the analysis and calculation steps in S3 specifically include: S301. Real-time compensation and correction of the corresponding state parameter data are performed using interference parameter data to eliminate errors caused by environmental fluctuations to the monitoring data.

[0099] Specifically, the interference parameter data synchronously collected in step S101 is used for real-time compensation and correction. Its implementation relies on deploying a separate, front-end data cleaning and compensation engine on a cloud or local data analysis platform. This engine, serving as a pre-processing module for the security assessment model, is centered around a library containing various compensation model algorithms.

[0100] When the platform receives a data packet from the aggregation point containing synchronously collected status parameters and interference parameters, it is first processed by the data cleaning and compensation engine. Based on the node identifier in the data packet header, the engine queries the compensation parameter configuration file bound to the node type (e.g., monitoring glass strain, steel structure vibration) and installation location (e.g., sunny or shady side). This configuration file pre-stores parameters related to the physical characteristics of the specific monitored object, such as the thermal expansion coefficient of curtain wall glass or aluminum alloy profiles, and the basic parameters of the vibration transfer function of the node structure to wind loads.

[0101] Furthermore, the specific software logic flow for compensation and correction is as follows: for each group of state and interference pairing data, the engine executes a standardized processing flow.

[0102] Taking the most common temperature-based strain data compensation as an example, the engine reads the original strain value ε1 and temperature value T1 from the current data, and at the same time reads a reference temperature T2 from the historical data cache of this node, which is usually the average temperature under stable conditions after installation or the temperature of the previous compensation cycle.

[0103] The compensation model is calculated according to the formula ε2=ε1-α*(T1-T2)*L, where α is the thermal expansion coefficient of the material and L is the sensitive length calibration coefficient. The calculated result ε2 is the mechanical strain after eliminating the thermal strain component caused by temperature.

[0104] For vibration data, the compensation model may be more complex. For example, when the amplitude of the low-frequency accelerometer data representing the overall wind-induced sway in the interference parameters exceeds the threshold, the engine will activate a digital filter, such as a band-stop filter, whose center frequency is dynamically adjusted according to the current wind speed. The current wind speed can be converted from the acceleration amplitude or obtained from an external meteorological data interface. It is used to filter out the energy in a specific frequency range caused by wind load from the original vibration spectrum and retain other frequency band signals that may represent structural loosening or damage.

[0105] All the above compensation calculations are completed before the data is formally sent into the core analysis submodule of the security assessment model, and a new corrected state parameter record with a timestamp and node ID is generated and stored in a dedicated database.

[0106] Based on this, the effects of harmful factors such as environmental disturbances are separated from the mixed effects through physical and mathematical models. The system uses the real-time monitored disturbance parameters as feedback signals to dynamically adjust the input of the compensation model and achieve adaptive correction.

[0107] S302. Perform a safety status determination based on the corrected status parameter data.

[0108] Therefore, in some implementations, environmental disturbances can be transformed from unquantifiable noise into known quantities that can be quantified and subtracted. When analyzing strain data, traditional methods struggle to determine whether a slowly rising trend indicates dangerous creep or a normal daily temperature cycle.

[0109] Through real-time temperature compensation in step S301, the system can accurately isolate the strain caused by temperature changes. If the remaining net mechanical strain trend still rises abnormally, it is very likely a real signal of structural problems, thus avoiding misreporting normal deformation caused by seasonal temperature differences as structural damage and reducing false alarms.

[0110] Conversely, in frigid weather, the actual risk to a structure may increase due to material embrittlement, but traditional raw data may show a decrease in strain values ​​due to cold contraction, masking the problem. Compensated data can more accurately reflect the stress state and avoid underreporting.

[0111] For vibration analysis, in strong winds, the original vibration signal can be drowned out by the loud wind noise, potentially making it impossible to detect genuine loosening or abnormal noises. By using dynamic filtering based on interference parameters such as wind-induced swaying signals, the system can effectively suppress wind noise background and improve the detection sensitivity of abnormal vibration characteristics.

[0112] Therefore, the safety status judgment performed in step S302, based on the corrected state parameter data, is decided upon by a purified and refined signal. This makes subsequent threshold comparisons, trend analysis, and pattern recognition more accurate and reliable, significantly enhancing the reliability and authority of the entire monitoring system's output conclusions. This enables the system to maintain stable high performance in complex and ever-changing outdoor environments, truly achieving reliable monitoring in all weather conditions.

[0113] In some embodiments, to significantly reduce reliance on cloud bandwidth and data analysis latency, near real-time alerts are achieved. The analysis and computation in step S3 are performed using a distributed edge collaborative computing architecture, specifically including: S303. Perform the first-level analysis at the data aggregation and communication gateway, filtering the data, extracting features, and quickly identifying anomalies.

[0114] Specifically, the implementation of Level 1 analytics relies on enhancements to the hardware architecture of data aggregation and communication gateways. These gateways are no longer simply protocol converters and data forwarders, but have been upgraded to edge intelligent gateways with certain edge computing capabilities.

[0115] In addition to traditional communication modules, its hardware core integrates a high-performance embedded application processor, adequate RAM, and non-volatile storage. A lightweight edge analytics engine software is deployed on this embedded processor. This engine incorporates the core algorithm library required for first-level analysis, including digital filters such as low-pass and band-pass filters, time-domain / frequency-domain feature extraction functions for calculating RMS, peak, and specific frequency band energy after FFT transformation, and one or more lightweight decision models for rapid anomaly detection, such as rule-based threshold comparators, or pre-trained and simplified miniature machine learning models, such as decision trees.

[0116] S304. Only when an anomaly is detected or at a fixed period, the processed feature data or anomaly event is uploaded to the cloud or local data analysis platform for second-level in-depth analysis and trend prediction.

[0117] Specifically, the software logic and data flow for the collaborative work of the first-level analysis in step S303 and the second-level in-depth analysis in step S304 are as follows: After the edge intelligent gateway receives the raw data packets from each monitoring node from its downstream wireless network, it first decrypts and parses them to extract the payload, namely the state parameters, interference parameters and node ID.

[0118] Subsequently, the edge analytics engine immediately performs pipelined processing on the data stream of each node independently: Filtering involves applying appropriate filters based on the node type to remove high-frequency noise or power frequency interference.

[0119] Feature extraction involves calculating a set of predefined key feature values ​​in real time from the filtered data. For example, for vibration data, the maximum amplitude and dominant frequency within the past 10 seconds can be extracted; for strain data, the current change relative to the reference value can be calculated.

[0120] Rapid anomaly detection involves inputting the extracted feature values ​​into a lightweight decision model. This model is designed for high sensitivity and low computational complexity, aiming to quickly identify "obvious anomalies" or "highly suspicious" events. Its output is a simple binary or ternary state label, such as "normal," "warning," or "alarm."

[0121] If the judgment result is "normal", the gateway usually only saves the feature value log of this processing in the local circular cache and does not immediately upload the original data.

[0122] The gateway will only trigger step S304 when the judgment result is "warning" or "alarm", or when a fixed long period such as every 6 hours is reached.

[0123] Once triggered, the gateway will upload the feature data packet related to this anomaly, including the node ID, timestamp, extracted feature vector, quick judgment result, or a structured anomaly event report, to the cloud platform via its uplink WAN link.

[0124] In this way, the overall data analysis task is spatially divided, with real-time and simple parts moved to the edge, and the timing and content of data uploads dynamically determined based on local analysis results, rather than a fixed full upload.

[0125] In summary, under the traditional architecture, all raw data must be transmitted to the cloud before analysis can begin. The combined effect of network transmission delay and cloud queuing delay results in a long lag time from the occurrence of an event to the generation of an alarm.

[0126] This implementation method decentralizes the first-level analysis to the gateway closest to the data source, enabling rapid judgment within hundreds of milliseconds to seconds after the data arrives at the gateway. Upon detecting an emergency anomaly, such as a severe impact vibration, the gateway can immediately send a tiny alarm event packet to the cloud via its uplink, and can also directly issue an on-site alarm through locally connected audible and visual alarms or by sending an SMS message, achieving near real-time response and saving valuable time for emergency handling.

[0127] Meanwhile, since the gateway only uses WAN bandwidth when uploading abnormal events or periodic summaries, and the uploaded data is refined feature data, the data volume is 1-2 orders of magnitude smaller than the original waveform data, which greatly reduces the continuous occupation of cloud bandwidth and cloud storage costs.

[0128] For cloud platforms, the data they receive has already undergone preliminary edge cleaning and labeling, avoiding the enormous computational pressure of processing massive amounts of raw data. This allows them to focus more on computationally intensive second-level in-depth analysis, such as trend prediction, complex pattern recognition, and multi-node correlation mining based on feature data spanning long periods of time.

[0129] Through a collaborative architecture that combines real-time edge filtering with deep insights in the cloud, the entire system can efficiently support the large-scale deployment of tens of thousands of monitoring nodes. While ensuring timely response to critical events, it achieves optimal allocation of system resources and a significant improvement in overall technical and economic efficiency.

[0130] In this embodiment, the method further includes: S4. Based on the assessment of the safety status, dynamically generate and output monitoring reports and early warning prompts for different locations or different risk levels.

[0131] Specifically, the dynamic generation and output process is implemented through a separate alarm and reporting service subsystem, which is closely integrated with the security assessment model.

[0132] When the S3 integrated decision-maker generates a new safety status judgment, especially when the status changes negatively, it immediately sends a structured event message to the service subsystem. This message includes at least the abnormal location described by the physical location mapped from the node's identity, such as the risk level (e.g., attention, warning, danger), the anomaly type (e.g., vibration exceeding limits, deformation acceleration), a timestamp, relevant data snapshots, and confidence level.

[0133] When the service subsystem receives an event, it first stores it in the event log database. Then, based on the preset risk level and response strategy rule base, it triggers the corresponding output action.

[0134] Furthermore, the output actions include various forms: Real-time alerts are provided. For alerts at the "Warning" level and above, the system automatically generates alert messages and sends them to designated operations and maintenance personnel or management platforms via SMS, in-app messages, email, etc., through integrated push notification services. While the alert messages are templated, key variables such as location, level, and type are dynamically populated to ensure clarity.

[0135] Dynamic monitoring reports are structured reports automatically generated by the system periodically or on demand, such as after an event occurs. These reports not only list event logs but also integrate historical data curves for relevant locations, data comparison charts with other nodes in the same area, risk trend analysis, and automatically highlight the highest-risk areas. The reports are presented as web pages or PDF documents and can be accessed through the platform or sent via email.

[0136] The visualization interface is updated so that the icon color representing different curtain wall areas or nodes on the graphical user interface of the monitoring system, such as a web-based GIS or BIM model, will change dynamically according to its current safety status, such as green, yellow, and red. Clicking on the icon allows you to drill down and view detailed data and event history.

[0137] Therefore, step S4 is a crucial step in transforming intelligent analysis results into practical action instructions, and its core advancement lies in the targeted and actionable nature of the output. Because the input event messages already contain judgments accurate to specific nodes and risk types, all output content possesses extremely high specificity.

[0138] Among them, the early warning message directly tells maintenance personnel "where to go", "what to look at" and "how urgent" the situation, that is, the precise location, risk type and risk level.

[0139] Dynamic reports provide detailed background data and trend analysis for decision-making, helping to determine whether immediate intervention is needed or whether to include the project in a monitoring plan.

[0140] The visual interface provides a global situational awareness, allowing managers to have a clear understanding of the safety heat map of the entire curtain wall at a glance.

[0141] This shift from general alarms to precise navigation significantly shortens the time from problem discovery to problem location and then to the development of a repair plan. It avoids blind, large-scale investigations, significantly improves the efficiency of maintenance work, reduces costs, and ensures that limited human resources can prioritize the most urgent and critical safety hazards, thereby maximizing the preventive safety protection value of the monitoring system.

[0142] In some embodiments, to reduce the risk of monitoring blind spots caused by equipment failure and improve the long-term stability and management efficiency of the system, the method further includes: S5. Periodically or trigger-based assessment of the data quality, signal strength, and energy level of each monitoring node.

[0143] Specifically, the performance evaluation of each monitoring node is implemented by deploying an independent network health monitoring engine in a cloud or local data analysis platform. This engine periodically or under specific event triggers, such as every hour or when a node has no data for three consecutive reporting cycles, proactively analyzes the operational status of each online monitoring node.

[0144] Its evaluation is based on three types of data sources: The metadata reported by the node, that is, in each data packet or periodic heartbeat packet, in addition to carrying sensor data, the node also includes health metadata such as its current battery voltage or supercapacitor voltage, internal chip temperature, and the signal strength (RSSI) and signal-to-noise ratio (SNR) of the most recent uplink measured by the communication module.

[0145] The platform-side communication gateway's link layer records include logs of each communication success / failure, retransmission count, and received signal quality with each node.

[0146] Application layer data quality analysis involves performing quality checks on the payloads reported by nodes, i.e., state parameter data. This includes checking data continuity (whether there is packet loss), data rationality (whether the values ​​are outside the physically possible extreme range), and data correlation (whether there are any contradictory abrupt changes with the data of neighboring nodes).

[0147] Furthermore, the specific software logic and algorithm for this assessment are as follows: the network health monitoring engine maintains a health state vector for each node. During each assessment cycle, the engine performs the following calculations: Regarding data quality, the algorithm calculates the effective reporting rate of the node's data within a recent time window, such as within 24 hours, which is the number of successful reporting cycles / the total number of cycles. It also combines the results of the rationality check, such as the number of data points marked as "invalid", to generate a quality score of 0-100.

[0148] For signal strength, the algorithm combines the uplink RSSI / SNR reported by the nodes and the downlink quality recorded by the gateway to calculate an average link quality index, and takes into account its stability, i.e., the fluctuation variance.

[0149] For energy levels, the algorithm estimates the remaining usable time or energy level by comparing the voltage value reported by the node with the discharge curve model of the node's battery.

[0150] These three metrics are compared to preset thresholds, such as a quality score < 80, average RSSI < -90 dBm, and estimated remaining days < 7 days. The system continuously monitors the status of its components and generates feedback information for self-maintenance.

[0151] In summary, traditional monitoring systems typically only focus on the status of monitored objects such as curtain walls, while remaining blind to the health status of the sensor network itself that performs the monitoring tasks. This is often only discovered passively after nodes have completely failed and data has been interrupted, resulting in monitoring blind spots of unknown duration.

[0152] This implementation method, through systematic multi-dimensional health assessment, endows the monitoring system with self-awareness capabilities. It can not only determine the safety of the curtain wall but also monitor in real time whether the monitoring tools themselves are functioning properly.

[0153] Specifically, by periodically and quantitatively assessing data quality, signal strength, and energy levels, the system can identify performance degradation trends in the early stages, before nodes completely fail.

[0154] For example, a slow decline in the signal strength of a node may indicate that its antenna is about to be completely covered by dirt or that its position has shifted slightly; a continuous drop in battery voltage is a clear warning of energy depletion.

[0155] This continuous insight into the state of the monitoring network itself is a step towards transforming passive fault response into proactive preventative maintenance, providing a data-driven basis for precise maintenance decisions in the S6 process.

[0156] S6. When the performance index of a certain node is found to be lower than the preset threshold, the node is automatically marked as requiring maintenance in the visualization results, and maintenance work order information is generated.

[0157] Specifically, the automatic tagging and generation of maintenance work orders relies on the deep integration of the network health monitoring engine with the system's alarm service subsystem and operation and maintenance management database. When the engine determines that any health indicator of a node, such as data quality, signal strength, or energy level, is below its preset threshold, it will immediately generate a standard-format node health anomaly event.

[0158] The event includes the following key fields: abnormal node ID; abnormal type, such as "insufficient energy," "weak signal," or "data anomaly"; specific numerical value of the abnormal indicator; trigger threshold; and detection timestamp. Subsequently, the event is pushed to the alarm service subsystem.

[0159] Furthermore, the software process for automatically generating maintenance information is as follows: after receiving a node health anomaly event, the alarm service subsystem first queries the system's node asset database. Based on the node ID in the event, it retrieves the detailed file of the node, including its installation location description, such as "South facade of the building, F5 floor, No. 3 vertical frame node"; node type; installation date; recent maintenance record, etc.

[0160] Next, the system calls the interface of the preset visualization rendering engine, and changes the color of the graphic element corresponding to the node to the preset "maintenance required" status color, such as flashing orange, on the graphical monitoring interface based on BIM or 2D elevation drawings. When the user clicks on the graphic element, a card pops up to display the specific health abnormality details.

[0161] Simultaneously, the alarm service subsystem automatically generates a structured preventative maintenance work order based on a predefined mapping rule base between anomaly types and maintenance actions. This work order includes at least the work order number; the location of the faulty node; recommended maintenance measures, such as "inspect and clean the node antenna and magnetic base," "replace the spare battery module," and "check the installation firmness and recalibrate"; the urgency level, dynamically categorized based on the anomaly type and the importance of the node's location; and a recommended optimal maintenance time window, for example, for nodes with insufficient energy, the system can predict "needs to be addressed within 3 days" based on their discharge rate.

[0162] In summary, this implementation method transforms health warnings into executable maintenance instructions, completing a closed loop from perception to action. In the traditional model, maintenance personnel need to conduct manual inspections or passively wait for fault reports, which is inefficient and carries a high risk of blind spots.

[0163] Through the S6 steps, the system can automatically and accurately locate abstract health indicator anomalies to specific, visualized physical locations and generate work orders with clear guidance. Maintenance personnel no longer need to perform tedious troubleshooting; they can simply take the work order and go to the designated location to perform the specified maintenance actions.

[0164] This significantly reduces the mean time to repair and transforms maintenance work from reactive emergency repairs to planned, predictable preventative operations. The system's consistently high online rate and data availability are thus guaranteed, effectively avoiding monitoring blind spots caused by the silent failure of individual nodes and ensuring the continuity and integrity of the entire curtain wall safety monitoring network data.

[0165] From a management perspective, automated work order generation and status tracking make the scheduling of operation and maintenance resources more scientific and reasonable, improve the efficiency and transparency of the overall operation and maintenance process, and significantly reduce the long-term operation and maintenance costs and management complexity of the system.

[0166] The following describes the safety monitoring system for building curtain walls provided in the embodiments of this application. The safety monitoring system for building curtain walls described below can be referred to in correspondence with the safety monitoring method for building curtain walls described above.

[0167] refer to Figure 2 The building curtain wall safety monitoring system is a three-tiered distributed hardware system, consisting of a physical layer (non-invasive sensor array 1), an edge layer (data aggregation and communication gateway 2), and a cloud layer (cloud or local data analysis platform 3), organically connected via wired and wireless networks. The system utilizes a three-tiered architecture of sensing, aggregation, and analysis to realize the entire process from data acquisition to intelligent decision-making. All hardware devices communicate via network protocols, ensuring smooth bidirectional data and control flow, jointly supporting the complete execution of the building curtain wall safety monitoring method.

[0168] The non-invasive sensor array 1 is deployed on the surface of the curtain wall and consists of multiple independent monitoring nodes. Each monitoring node includes a sensing module for collecting physical state parameters of the curtain wall and a connection interface for physically connecting with the curtain wall surface.

[0169] The sensing module, serving as the data source, is primarily composed of multiple miniaturized sensor chips soldered onto a circuit board inside the node. These mainly include a triaxial MEMS accelerometer chip for measuring the curtain wall's vibration acceleration and static tilt angle changes; a high-precision digital temperature sensor chip, mounted close to the inside of the node's outer shell, for measuring the surface temperature at the node's location, serving as a basis for environmental compensation of state parameters; and a micro-strain sensing unit, which can employ a microelectromechanical system strain sensor or indirectly sense stress by measuring the micro-deformation of the flexible base in contact with the curtain wall.

[0170] The connection interface adopts a reversible connection method based on adsorption or magnetism, which allows the entire monitoring node to be installed and removed without damaging the curtain wall structure.

[0171] The flexible vacuum adsorption method involves molding multiple highly elastic silicone micro-suction cups at the bottom of the node. The inner cavity of the suction cup is connected to a micro-piezoelectric ceramic vacuum pump through a microchannel. The pump's start and stop are controlled by a microcontroller to generate or release negative pressure adsorption force. The electromagnetic adsorption method involves embedding a magnetic array composed of multiple small neodymium iron boron magnets arranged in a specific polarity at the bottom of the node, or installing a planar electromagnet array inside the node.

[0172] For the latter, a thin magnetic conductive sheet, such as a flexible silicon steel sheet, needs to be pre-attached to a predetermined position on the curtain wall surface. This sheet generates a strong magnetic force when energized, and the magnetic force essentially disappears when the power is off. Both interfaces ensure that the nodes can be installed by hand, firmly pressed and adhered, and also removed by hand.

[0173] During installation, the node is attached to the designated measurement point on the curtain wall via its connection interface. After the node is powered on, the microcontroller wakes up according to a preset cycle, drives the sensor module to collect data, and simultaneously reads the identification code from the memory. The two are packaged together, and the data packet is sent through the wireless communication unit. In Mesh network mode, the node's wireless communication module hardware supports routing functions, and the corresponding protocol stack runs within the microcontroller, enabling it to forward data from neighboring nodes.

[0174] Data aggregation and communication gateway 2 is a fixed edge computing hardware device deployed inside a building. It is used to receive monitoring data from a non-intrusive sensor array and perform preliminary data packaging and protocol conversion.

[0175] The gateway employs a high-performance embedded processor and is equipped with hundreds of megabytes to gigabytes of RAM and Flash memory, providing the necessary computing and storage resources for the edge analytics engine software running on it.

[0176] This gateway features a multi-mode communication interface, comprising a downlink communication module and an uplink communication module. The downlink communication module is equipped with a corresponding concentrator or coordinator hardware module depending on the wireless protocol used by the sensor array. The uplink communication module comes standard with an Ethernet interface and can be optionally equipped with a 4G / 5G module or a fiber optic module for internet access and uploading data to a cloud platform.

[0177] The gateway continuously listens for and receives data packets from the sensor array via its downlink communication module. The software running on the edge computing core processes the raw data in real time, including digital filtering, time-frequency domain feature extraction, and running a lightweight, fast anomaly detection model.

[0178] Based on the assessment results, a decision is made as to whether to record data locally only or upload the feature data or event report to the cloud via the uplink communication module. This gateway is also responsible for managing the topology, routing, and maintenance of the Mesh network.

[0179] Cloud-based or local data analysis platform 3, which receives data from data aggregation and communication gateways, executes security status analysis algorithms, and generates visualization results and early warning information.

[0180] Specifically, the platform's hardware infrastructure includes: The access layer server is responsible for load balancing and securely receiving uplink data connections from all gateways across the network.

[0181] A computing server cluster consists of multiple high-performance application servers, possibly equipped with GPUs to accelerate machine learning inference. These servers house a data cleaning and compensation engine that corrects raw data using interference parameters such as temperature; a core security assessment model for deep analysis, trend prediction, and pattern recognition; a network health monitoring engine to evaluate node performance; and an alarm and reporting service to generate alerts and work orders.

[0182] A data storage server cluster is a cluster of servers that deploy time-series databases, relational databases, etc., for persistent storage of massive amounts of monitoring data, node files, event logs, system configuration parameters, etc.

[0183] Web / application servers provide API interfaces and graphical web interfaces to present visualized information such as analysis results, early warning maps, and maintenance work orders to end users.

[0184] After receiving data from the gateway, the platform's data cleaning and compensation engine first performs preprocessing. Then, the security assessment model analyzes the processed data and historical data to draw security status conclusions, while the network health monitoring engine analyzes node health indicators in parallel.

[0185] Once a curtain wall safety anomaly or node health issue is detected, the alarm and reporting service is immediately triggered, generating an early warning message and updating the visualization interface, while automatically creating a maintenance work order. All process data and results are stored in the database cluster.

[0186] Each monitoring node in the non-invasive sensor array has an independent identification and is bound to a specific physical location on the curtain wall surface.

[0187] In summary, this system employs a non-invasive sensor array constructed using reversible connection interfaces based on adsorption or magnetism, completely eliminating traditional destructive installation methods such as drilling and bonding. While achieving full-surface, location-based monitoring, it absolutely guarantees the integrity and aesthetics of the curtain wall structure. The system architecture clearly defines the three levels of sensing, aggregation, and analysis, and introduces a node identity and location binding mechanism, providing a hardware foundation for high-precision, traceable monitoring and significantly improving the system's deployment versatility, maintainability, and applicability to historical buildings.

[0188] In some embodiments, each monitoring node in the non-invasive sensor array 1 further includes a micro-power management unit, which is a dedicated power subsystem integrated on the internal circuit board of each monitoring node to achieve energy autonomy of the node.

[0189] Specifically, the micro-energy management unit integrates an environmental energy harvesting device, a rechargeable energy storage element, and an intelligent power management circuit.

[0190] Among them, the environmental energy harvesting device is a multi-source input energy harvesting front end, which typically includes one or more of the following transducers: The photovoltaic collection module, employing ultra-thin, flexible amorphous silicon or dye-sensitized solar cells, is attached to the upper surface or side of the node's outer shell beneath a transparent window. Its area is optimized to generate a small current under both indoor lighting and outdoor diffused light conditions. In this embodiment, the module is designed to cover a portion of the node's outer shell's curved surface to accommodate light reception at different installation angles.

[0191] The thermoelectric collection module consists of multiple miniature thermoelectric generators connected in series or parallel. These generators are installed inside the node's outer shell, with their hot ends in close contact with the inner wall of the shell via highly thermally conductive silicone grease to sense the temperature of the curtain wall surface; their cold ends face the air inside the node or are connected to a small heat sink. Utilizing the fluctuating temperature difference on the curtain wall surface caused by sunlight, day-night cycles, or indoor-outdoor temperature differences, the Seebeck effect generates a voltage on the thermoelectric generator.

[0192] The piezoelectric / electromagnetic vibration collection module uses a combination of piezoelectric ceramic sheets or miniature electromagnetic induction coils and magnets. The piezoelectric sheet is typically attached to a specific location on the node circuit board or connected to the inside of the casing. When the curtain wall or node experiences minor mechanical vibrations due to wind or environmental vibrations, the piezoelectric material deforms and generates an electric charge. The electromagnetic type utilizes the relative motion between the magnet and the coil to cut magnetic field lines and generate current.

[0193] The environmental energy harvesting device is configured to harvest energy from the light, temperature difference or vibration of the environment in which the monitoring node is located, and to power the sensing module and communication components, thereby enabling the long-term self-sustaining operation of the monitoring node.

[0194] Furthermore, rechargeable energy storage elements typically employ thin, small-volume lithium polymer batteries or supercapacitors, which are better suited for frequent charge-discharge cycles. This element is soldered onto a circuit board and is responsible for storing the electrical energy generated by the harvesting device and powering the entire node when the harvesting device's power is insufficient.

[0195] Furthermore, the intelligent power management circuit is a dedicated power management chip and its peripheral circuitry, which includes: Energy input management involves rectifying and boosting the unstable and weak AC or DC power generated by different collection devices to stabilize it at a voltage level that can charge energy storage components.

[0196] Multi-source prioritization and maximum power point tracking can intelligently select the main energy source according to environmental conditions. For example, photovoltaics are used first during the day, while thermoelectricity or vibration is relied on at night, and the operating point is dynamically adjusted to extract the maximum power from each collection device.

[0197] Energy storage management is the safe control of charging energy storage devices and the prevention of overcharging and over-discharging.

[0198] Dynamic voltage regulation and load management provide different levels of stable voltage based on the operating status of the sensing modules, microcontrollers, and communication modules within the node, and precisely control the power supply to each part to achieve ultimate energy saving.

[0199] Based on this, the operation of the micro-energy management unit is a closed loop of continuous energy harvesting, storage, and supply. First, photovoltaic, thermoelectric, and piezoelectric devices continuously convert light, heat, and mechanical energy in the environment into electrical energy at the microwatt to milliwatt level.

[0200] Secondly, the power management chip receives this electrical energy, conditions it, and charges the rechargeable energy storage element with appropriate current and voltage. The chip constantly monitors the voltage state of the energy storage element.

[0201] Finally, when the node microcontroller needs to operate, it sends a request to the power management chip. The chip determines whether to wake up the main system based on the charge level of the energy storage components. During power supply, the chip converts the voltage of the energy storage components to a stable 3.3V or 1.8V, etc., to supply data to the sensing module for data acquisition, the microcontroller for data processing, and the communication module for data transmission. After one working cycle is completed, the power management chip cuts off the high-voltage power supply to most circuits, maintaining only its own and the monitoring circuits of the energy storage components operating at extremely low power consumption, awaiting the next acquisition cycle or an external wake-up signal.

[0202] In summary, by integrating a micro-energy management unit into each non-intrusive node, and especially by combining various environmental energy harvesting technologies, the problem of continuous power supply for passively deployed nodes is fundamentally solved. This frees the monitoring nodes from dependence on fixed power sources or frequent battery replacements, enabling them to operate immediately upon deployment. This greatly expands the system's application capabilities in scenarios without external power sources and further reduces the complexity and cost of operation and maintenance throughout the entire lifecycle.

[0203] This application provides an electronic device, such as... Figure 3 As shown, Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3The illustrated electronic device 300 includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302. Optionally, the electronic device 300 may also include a transceiver 304. It should be noted that in practical applications, the transceiver 304 is not limited to one type, and the structure of this electronic device 300 does not constitute a limitation on the embodiments of this application.

[0204] Processor 301 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in connection with the embodiments of this application. Processor 301 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0205] Bus 302 may include a pathway for transmitting information between the aforementioned components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 302 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0206] The memory 303 may be a ROM (Read-Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or it may be an EEPROM (Electrically Erasable Programmable Read-Only Memory), a CD-ROM (Compact Disc Read-Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0207] The memory 303 is used to store application code that executes the scheme of the embodiments of this application, and its execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the content shown in the foregoing method embodiments.

[0208] Among them, electronic devices include, but are not limited to: mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments described in this application.

[0209] This application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the steps of the safety monitoring method for building curtain walls described above.

[0210] Since the embodiments of the computer-readable storage medium portion correspond to the embodiments of the method portion, please refer to the description of the embodiments of the method portion for the embodiments of the computer-readable storage medium portion.

[0211] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0212] The above are only some embodiments of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A building curtain wall safety monitoring system, characterized in that, include: The non-invasive sensor array consists of multiple independent monitoring nodes. Each monitoring node includes a sensing module for collecting physical state parameters of the curtain wall and a connection interface for physically connecting with the curtain wall surface. The connection interface adopts a reversible connection method based on adsorption or magnetism, which allows the entire monitoring node to be installed and removed without damaging the curtain wall structure. A data aggregation and communication gateway is used to receive monitoring data from the non-invasive sensor array and to perform preliminary data packaging and protocol conversion. And a cloud or local data analysis platform, used to receive data from the data aggregation and communication gateway, execute security status analysis algorithms, and generate visualization results and early warning information; Each monitoring node in the non-invasive sensor array has an independent identification and is bound to a specific physical location on the curtain wall surface.

2. The system according to claim 1, characterized in that: Each monitoring node in the non-invasive sensor array also includes a micro-energy management unit, which integrates an environmental energy harvesting device and a rechargeable energy storage element. The environmental energy harvesting device is configured to collect energy from the light, temperature difference, or vibration of the environment in which the monitoring node is located, and to power the sensing module and communication components, thereby enabling the long-term self-sustaining operation of the monitoring node.

3. The system according to claim 1, characterized in that: The connection interface specifically includes a flexible vacuum suction cup array or a programmable electromagnetic adsorption module; wherein, the flexible vacuum suction cup array can adaptively conform to the curved or uneven surface of the curtain wall to form a sealed negative pressure zone, and the programmable electromagnetic adsorption module can generate a strong adsorption force when powered on and completely release when powered off.

4. A method for monitoring the safety of building curtain walls, characterized in that, Applied to the system as described in any one of claims 1-3, the method comprises: Multiple monitoring nodes are reversibly attached to the surface of the curtain wall to collect state parameter data of the curtain wall at different locations in parallel. The status parameter data collected by each monitoring node, along with its identification information, are aggregated to the data aggregation point via a wireless communication network. Based on a pre-set safety assessment model, the aggregated status parameter data with location markers are analyzed and calculated to determine the overall and local safety status of the curtain wall. Based on the assessment of the safety status, monitoring reports and early warning prompts are dynamically generated and output for different locations or different risk levels.

5. The method according to claim 4, characterized in that, The parallel acquisition step also includes: When collecting the status parameter data, each monitoring node simultaneously collects the interference parameter data of its local environment.

6. The method according to claim 5, characterized in that, The analysis and calculation steps specifically include: The interference parameter data is used to compensate and correct the corresponding state parameter data in real time, so as to eliminate the error caused by environmental fluctuations to the monitoring data. Perform a safety status assessment based on the corrected status parameter data.

7. The method according to claim 4, characterized in that, The step of summarizing data to the data aggregation point specifically includes: A multi-hop ad hoc network communication protocol is adopted, in which some monitoring nodes are configured as routing nodes, which can forward data from other adjacent monitoring nodes, thereby constructing a mesh data transmission network with redundant paths covering the entire curtain wall area within the wireless communication network.

8. The method according to claim 4, characterized in that, The method further includes: Regularly or trigger-based assessments of the data quality, signal strength, and energy levels of each monitoring node; When a node's performance metric is found to be below a preset threshold, the node is automatically marked as requiring maintenance in the visualization results, and a maintenance work order is generated.

9. The method according to claim 4, characterized in that, The analysis and calculation steps are executed using a distributed edge collaborative computing architecture, specifically including: The first-level analysis is performed at the data aggregation and communication gateway, which includes data filtering, feature extraction, and rapid anomaly detection. Only when an anomaly is detected or at a fixed period, the processed feature data or anomaly events are uploaded to the cloud or local data analysis platform for second-level in-depth analysis and trend prediction.

Images