Remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing

Abstract

The present application relates to the technical field of construction intelligent monitoring, in particular to a building construction remote intelligent monitoring system based on 5G telephone communication and edge computing, comprising: a field sensing unit; an edge processing unit, which realizes accurate integration of heterogeneous data through an improved adaptive weighted fusion algorithm, and generates real-time control strategies in combination with a working condition dynamic model; a 5G communication unit, which combines predictive slice scheduling and a multi-modal anti-interference protocol; and a remote monitoring unit.The present application uses an industrial-grade sensor array to collect relevant data and generate monitoring data streams with timestamps and abnormality markers, and the edge processing unit completes abnormal data through a data preprocessing module and realizes heterogeneous data integration through an improved adaptive weighted fusion algorithm, effectively solving the problems of insufficient adaptation of data processing architecture to construction multi-source data characteristics and missing data effectiveness traceability, and realizing effective traceability and accurate integration of multi-dimensional monitoring data in a construction scene.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction monitoring technology, and more specifically, to a remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing. Background Technology

[0002] Construction scenarios require real-time monitoring of personnel locations, equipment operating parameters, and environmental physical parameters to ensure safety and efficiency. Traditional monitoring solutions suffer from limited data acquisition dimensions, high processing latency, and poor stability of remote transmission. The low latency of 5G communication and the proximity data processing capabilities of edge computing offer a direction for solving these problems. However, how to achieve accurate fusion of multi-source heterogeneous data, reliable transmission of high-priority data, and control strategies that dynamically adapt to working conditions remains a key bottleneck that needs to be overcome.

[0003] In existing technologies, relevant patents have explored the application of sensor acquisition, 5G transmission, and edge technology in the field of engineering monitoring. For example, Chinese patent CN202510436294.0 discloses a 5G-based intelligent pipeline inspection and remote monitoring system and method based on edge IoT. By deploying sensors and high-definition cameras in the communication pipeline to collect data, and after 5G transmission, edge technology is used to denoise and standardize the data and fuse multi-sensor features. Combined with anomaly detection, the health status of the pipeline is assessed to achieve real-time monitoring and intelligent inspection of the pipeline. Another example is Chinese patent CN202311138111.4, which discloses a 5G communication-based engineering construction monitoring system, including a 5G communication device, a security module, a construction monitoring module, and an equipment monitoring module. It can collect the physical condition of workers through smart bracelets, avoiding the inability to report or send medical attention in a timely manner during emergencies.

[0004] Despite the advantages of the aforementioned technical solutions, the following technical shortcomings remain: First, the data processing architecture is not well adapted to the multi-source data characteristics of construction scenarios, and a data validity traceability mechanism is lacking. For example, Chinese patent CN202510436294.0, designed for communication pipeline inspection scenarios, does not consider the collaborative processing needs of multi-dimensional data from personnel, equipment, and the environment during construction. Its data processing focuses only on noise reduction and standardization, failing to label data anomalies for common construction scenarios such as sensor power supply fluctuations and signal packet loss, and also failing to directly bind timestamps to data collection actions, making it impossible to trace the time consistency and validity of data collection. Similarly, Chinese patent CN202311138111.4 does not introduce an edge computing architecture, requiring all data to be transmitted to a remote end for processing, which is prone to processing delays due to transmission distance. Furthermore, it does not record abnormal states and time-related information of the collected data, making it difficult to support accurate data analysis and decision-making in construction scenarios. Secondly, communication transmission lacks dynamic adaptability and multi-service collaboration mechanisms for construction scenarios: Chinese patent CN202510436294.0's 5G transmission only serves as a basic data channel, without designing a "scenario-awareness-resource adaptation" mechanism, lacking predictive slicing scheduling to cope with bandwidth demand fluctuations caused by changes in construction intensity, and lacking a multi-modal anti-interference protocol to handle electromagnetic interference and signal obstruction issues in construction areas, thus failing to guarantee ultra-low latency transmission of high-priority data such as equipment operating parameters; Chinese patent CN202311138111.4's 5G communication only realizes basic interaction of data and personnel information, without considering seamless collaboration between voice communication and data transmission, and without dynamically adjusting transmission resources according to channel parameters (such as signal strength and bit error rate) in construction areas, making it difficult to adapt to the dynamically changing channel characteristics in construction scenarios. Third, the control strategies have limited coverage and lack tiered response capabilities to match construction conditions: The control strategy of Chinese patent CN202510436294.0 only generates inspection and adjustment instructions based on pipeline health status, failing to cover multi-dimensional management needs such as personnel safety, equipment malfunctions (e.g., excessive vibration, high temperature), and environmental parameter warnings (e.g., excessive dust, excessive noise) during construction. The control strategy of Chinese patent CN202311138111.4 only focuses on warnings of workers' physical condition, lacking response logic for equipment failures and environmental anomalies, and failing to generate tiered control instructions (e.g., parameter fine-tuning, equipment shutdown, personnel evacuation) based on the severity of the anomaly, making it difficult to cope with the complex and diverse risks in construction scenarios. Therefore, we propose a remote intelligent monitoring system for construction based on 5G telephone communication and edge computing. Summary of the Invention

[0005] The purpose of this invention is to provide a remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing, in order to solve the problems mentioned in the background art, such as insufficient adaptation of the data processing architecture to the multi-source data characteristics of construction scenarios, lack of data validity traceability mechanism, lack of dynamic adaptation capability of communication transmission to construction scenarios, limited coverage of multi-service collaboration mechanism and control strategy, and lack of hierarchical response capability matched with construction conditions.

[0006] To address the aforementioned technical problems, the present invention aims to provide a remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing, comprising:

[0007] The field sensing unit uses an industrial-grade sensor array to collect data on the location of construction personnel, equipment operating parameters, and environmental physical parameters, generating a monitoring data stream with timestamps.

[0008] An edge processing unit is communicatively connected to a field sensing unit. It constructs a multi-source data processing engine based on an edge computing architecture, achieves accurate integration of heterogeneous data through an improved adaptive weighted fusion algorithm, and generates real-time control strategies by combining the dynamic working condition model.

[0009] The 5G communication unit is connected to the edge processing unit and the remote monitoring unit respectively. Through the "scene perception-resource adaptation" dual closed-loop transmission mechanism, combined with predictive slicing scheduling and multimodal anti-interference protocol, it realizes the ultra-low latency transmission of high-priority data and seamless coordination of voice communication, adapting to the dynamic channel characteristics of the construction scenario.

[0010] The remote monitoring unit receives edge processing results through a 5G communication unit, stores historical data and displays real-time status, and sends parameter adjustment instructions to the edge processing unit to form a closed-loop monitoring link.

[0011] As a further improvement to this technical solution, the industrial-grade sensor array of the field sensing unit includes a personnel positioning module, an equipment sensing module, and an environmental monitoring module, wherein:

[0012] The personnel positioning module uses a Beidou positioning sensor, which is integrated inside the smart safety helmet worn by construction workers. The sensor is connected to the built-in microcontroller of the safety helmet through an industrial standard serial interface. It collects the three-dimensional position coordinates of construction workers at a frequency that meets the safety tracking needs of construction workers. During the collection process, the power consumption of the sensor meets the industrial low-power equipment operation standards.

[0013] The equipment sensing module includes a vibration sensor and a temperature sensor. The vibration sensor is fixed to the surface of the load-bearing component of the key construction equipment (such as tower crane motor, construction elevator drive shaft, etc.), and the temperature sensor is attached to the heat-generating component inside the equipment distribution box. Both the vibration sensor and the temperature sensor collect data at a frequency that is adapted to the monitoring requirements of the equipment's operating status. The output signals of the vibration sensor and the temperature sensor are transmitted after analog-to-digital conversion.

[0014] The environmental monitoring module uses dust sensors, noise sensors, and temperature and humidity sensors, which are distributed according to the spacing of the key construction areas (material storage area, scaffolding perimeter). The dust sensors, noise sensors, and temperature and humidity sensors are connected to the main controller of the field sensing unit through an industrial bus, and collect data at a frequency that adapts to the monitoring needs of environmental parameter changes. The sensor housing is designed with a protection level suitable for construction dust and humid environments.

[0015] As a further improvement to this technical solution, the field sensing unit generates a timestamped monitoring data stream, specifically including:

[0016] With the built-in verification function of the industrial-grade sensor array, when collecting the location of construction personnel, equipment operating parameters and environmental physical parameters, the power supply stability and signal transmission integrity of each sensor are monitored simultaneously. When the power supply of the sensor exceeds its normal operating range or when packet loss occurs in the signal transmission, an abnormality mark is automatically marked next to the timestamp of the corresponding collected data.

[0017] By binding the time stamp generation moment to the actual sensor acquisition action, the time stamp is generated synchronously at the moment the sensor triggers acquisition, avoiding time deviation caused by data temporary storage and ensuring the consistency of the time stamp and parameter acquisition behavior in the time dimension.

[0018] The generated monitoring data streams are distinguished and labeled according to parameter types (construction personnel location, equipment operating parameters, environmental physical parameters). The labeling information is embedded in the data stream header, enabling the edge processing unit to directly identify the data category and match the corresponding processing strategy.

[0019] As a further improvement to this technical solution, the edge processing unit includes a data preprocessing module, a feature extraction module, and a fusion decision module, wherein:

[0020] The data preprocessing module receives the time-stamped monitoring data stream output by the field sensing unit, performs time-domain smoothing on the data through a sliding window to filter out high-frequency interference signals, completes the data marked as abnormal (including the abnormal identifiers marked by the field sensing unit) using an interpolation algorithm based on the distribution law of data from the same type of sensor during the same period, and physically truncates extreme data that exceed the sensor's range (the truncation value is set based on the sensor's rated operating range). During the processing, the one-to-one correspondence between the data and the original timestamp is maintained.

[0021] The feature extraction module extracts dimensional features from the preprocessed data. It extracts the direction vector of the movement trajectory and the regional stay feature from the construction personnel location data, extracts the steady-state component and transient fluctuation feature of the operating state from the equipment operating parameters, and extracts the gradient feature and cumulative effect feature of parameter changes from the environmental physical parameters. All extracted features are output in the form of standardized numerical vectors, with the vector dimension and data type corresponding one-to-one.

[0022] The fusion decision module aligns the multi-dimensional feature vectors output by the feature extraction module with spatial coordinates, assigns fusion coefficients to different features through the feature contribution evaluation model, and integrates the weighted feature vectors into a unified working condition evaluation matrix. The matrix dimension adapts to the multi-source data processing engine calculation requirements of the edge processing unit.

[0023] As a further improvement to this technical solution, the edge processing unit achieves accurate integration of heterogeneous data through an improved adaptive weighted fusion algorithm, including the following steps:

[0024] S240.1 Receive multi-source heterogeneous data processed by the data preprocessing module. The data includes construction personnel location data, equipment operating parameters, and environmental physical parameters. Assign a unique data source identifier to each type of data and initialize the real-time reliability coefficient of each data source. Initial value;

[0025] S240.2 Calculate the real-time reliability coefficient of each data source based on the data source identifier. By analyzing the temporal consistency (the range of fluctuation in continuous sampled values) and spatial correlation (the degree of deviation of data from similar sensors in the same region), the data can be... Mapped to the [0,1] range, the higher the value, the higher the reliability of the data source;

[0026] S240.3, according to Dynamically allocate fusion weights Satisfying the weight constraint relationship When a certain data source When the reliability falls below a preset threshold, the corresponding data source will be automatically downgraded. The percentage, the released weight share, is determined by other highly reliable data sources according to... Proportional sharing;

[0027] S240.4. Data is processed using a hierarchical fusion method, first through a local fusion operator. Data from similar sensors is fused (e.g., vibration parameter fusion from multiple devices, environmental parameter fusion from multiple regions), and then fused using a global fusion operator. The results of different types of local fusion are integrated to generate a global fusion result. ;

[0028] S240.5, Through deviation verification factor verify The effectiveness, Based on the statistical patterns of deviations between historical fusion data and actual working conditions, if... The deviation exceeds If the range is not cleared, return to S240.3 to readjust the weight allocation until a valid fusion result is generated.

[0029] As a further improvement to this technical solution, the edge processing unit generates a real-time control strategy by combining the dynamic working condition model, including the following steps:

[0030] S250.1, Recall the data stored in the operating condition feature library. Typical construction condition standard feature vector The standard feature vector Generated by clustering historical normal operating condition data, each standard feature vector contains equipment operation feature components, personnel distribution feature components, and environmental parameter feature components;

[0031] S250.2 Extract the global fusion result after processing with the improved adaptive weighted fusion algorithm. Corresponding real-time feature vector Through the vector space distance function calculate With each standard eigenvector similarity distance Select The three smallest standard operating conditions are selected as the current matching candidate set;

[0032] S250.3. Based on the distribution patterns of historical abnormal data of candidate operating conditions, generate dynamic thresholds for the current operating conditions. , The correction amount is positively correlated with the historical anomaly deviation of the candidate set operating conditions, and the corrected amount is... This includes equipment safe operation threshold ranges, personnel safe density thresholds, and environmental parameter early warning thresholds;

[0033] S250.4, Combine the global fusion results With dynamic threshold When a comparison is performed, When a certain parameter exceeds the corresponding threshold range, a graded control strategy is generated based on the degree of deviation: when the deviation is slight, a parameter adjustment instruction is generated (such as fine-tuning of equipment operating frequency or starting of environmental control equipment); when the deviation is severe, an emergency intervention instruction is generated (such as equipment shutdown instruction or personnel evacuation warning), and the control strategy is encapsulated and transmitted to the 5G communication unit.

[0034] As a further improvement to this technical solution, the 5G communication unit includes a communication interface module and a link monitoring module, wherein:

[0035] The communication interface module establishes connections with the edge processing unit and the remote monitoring unit respectively. It connects to the edge processing unit via an industrial Ethernet interface (compliant with the IEEE 802.3 standard), transmitting data through shielded twisted-pair cables. It also integrates NTP time synchronization to maintain consistency with the clock reference of the edge processing unit, ensuring that timestamped monitoring data streams, global fusion results, and parameter adjustment commands are transmitted without time deviation. It connects to the remote monitoring unit via a 5G NR wireless interface (compliant with 3G PPR15 and above standards), supporting independent networking mode. The interface transmission capacity adapts to the bandwidth requirements of the "scene awareness-resource adaptation" dual closed-loop transmission mechanism, providing a basic transmission link for high-priority data and voice communication.

[0036] The link monitoring module collects link status parameters of the communication interface module in real time, including data transmission rate, bit error rate, and signal strength. When the link bit error rate exceeds the transmission reliability threshold for the adapted construction scenario, or the signal strength is lower than the critical value for ensuring continuous data transmission, a link abnormality signal is generated and fed back to the "scenario perception-resource adaptation" dual closed-loop transmission mechanism. This provides a link status basis for resource adaptation adjustment, ensuring dynamic matching between link status and resource allocation, and adapting to the dynamic channel characteristics of the construction scenario.

[0037] As a further improvement to this technical solution, the 5G communication unit also includes a scene perception module and a resource adaptation module, wherein:

[0038] The scene perception module receives the working condition feature vector output by the edge processing unit in real time, and simultaneously collects the 5G channel parameters of the construction area. It then integrates the working condition feature vector and the channel parameters by timestamp to generate a scene perception matrix. ,and The time dimension and the global fusion result of the edge processing unit output Maintaining synchronization provides spatiotemporally consistent scenario data support for resource adaptation; among which As a dimension of working condition characteristics, For channel parameters;

[0039] The resource adaptation module is based on a scene-aware matrix. Perform predictive slice scheduling when medium construction intensity At that time, a dedicated network slice is allocated to high-priority data such as equipment operating parameters. ,in The threshold for determining high-intensity work; when Medium signal blocking coefficient At the same time, dynamically expand the slice bandwidth and adapt to the modulation and coding scheme. ( With Doppler frequency shift (Adjusting the order of changes) ensures that network slice resources are matched with scenario requirements in real time, enabling ultra-low latency transmission of high-priority data (ensuring latency). Not exceeding the construction safety response delay threshold );in This is the critical value for signal attenuation.

[0040] As a further improvement to this technical solution, the 5G communication unit also includes an anti-interference processing module and a voice-data coordination module, wherein:

[0041] The anti-interference processing module operates based on a multi-modal anti-interference protocol and has a built-in channel monitoring component that collects the electromagnetic interference intensity of the construction area in real time. With multipath effect parameters When sudden electromagnetic interference is detected ( Reaching the interference tolerance threshold When, based on the detected interference frequency band Generate frequency hopping pattern and set frequency hopping interval. When signal obstruction in a fixed area is detected ( Reaching the signal attenuation threshold When ), 3D beamforming technology is activated, through angle Adjust signal propagation direction (related to the height of obstructions) With transmission distance (Diffraction of obstructions to ensure low bit error rate in signal transmission) Not exceeding the industrial-grade transmission reliability threshold ;

[0042] The voice-data collaboration module achieves seamless collaboration between voice communication and high-priority data through 5G VoNR technology, mapping voice signals to QoS streams. (Guarantee delay) Mapping high-priority data to QoS streams (Guaranteed bit rate) ); adopts a transmission frame structure of "voice frame embedded data subframe", each A high-priority data subframe is embedded after each consecutive voice frame, and the timestamp of the data subframe is aligned with that of the voice frame to ensure uninterrupted voice communication and zero data transmission delay, achieving collaborative adaptation between the two. This is the frame synchronization coefficient, adapted to the speech sampling rate.

[0043] As a further improvement to this technical solution, the remote monitoring unit includes a data management module, a visualization interaction module, and a command-analysis closed-loop module, wherein:

[0044] The data management module employs a hierarchical storage architecture to process edge processing results and real-time global fusion results. eigenvectors Millisecond-level read and write operations are performed using an in-memory database to ensure real-time data access efficiency; historical operating data is stored in a columnar database partitioned by time dimension, supporting fast retrieval of TB-level data, and the storage strategy is dynamically adapted to the transmission bandwidth of the 5G communication unit.

[0045] The visualization interaction module is used to build a digital twin visualization interface for the construction scene, which transforms the edge processing results into dynamic identification of equipment status, dynamic distribution trajectory of personnel and real-time curves of environmental parameters. It supports synchronous refresh of the interface on multiple terminals, and the refresh frequency is consistent with the global fusion cycle of the edge processing unit.

[0046] The instruction-analysis closed-loop module realizes the encrypted issuance of parameter adjustment instructions and historical data-driven analysis: the instructions issued to the edge processing unit are encrypted using the national cryptographic SM4 algorithm and digitally signed, and the entire instruction process is recorded through blockchain notarization technology; at the same time, construction trend analysis is performed based on historical data, and analysis results are generated to provide data support for instruction generation, forming a closed-loop link of "instruction issuance - result feedback - trend analysis - instruction optimization".

[0047] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0048] 1. This invention uses an industrial-grade sensor array in the field sensing unit to collect the location of construction personnel, equipment operating parameters, and environmental physical parameters, and generates a monitoring data stream with timestamps and anomaly markers (power fluctuations, signal packet loss). At the same time, the edge processing unit completes the abnormal data through the data preprocessing module and uses an improved adaptive weighted fusion algorithm (dynamically allocating weights based on the real-time reliability coefficient of the data source and fusion of the same and different types of data in a layered manner) to achieve heterogeneous data integration. This effectively solves the problems of insufficient adaptation of the data processing architecture to the characteristics of multi-source construction data and lack of data validity traceability, and realizes effective traceability and accurate integration of multi-dimensional monitoring data in the construction scenario.

[0049] 2. This invention generates a scene perception matrix through the scene perception module of the 5G communication unit. The resource adaptation module performs predictive slice scheduling based on the matrix (such as allocating dedicated slices for equipment parameters when the construction intensity meets the standard, expanding bandwidth and adjusting the modulation and coding scheme when the signal is blocked). The anti-interference processing module adopts a multi-modal anti-interference protocol (frequency hopping when there is sudden interference, and enabling 3D beamforming when there is fixed blockage). The voice-data collaboration module realizes frame-embedded transmission of voice and high-priority data through 5G VoNR technology, which effectively solves the problem of lack of dynamic adaptation capability and multi-service collaboration mechanism for communication transmission in construction scenarios, and ensures the ultra-low latency transmission of high-priority data and seamless collaboration of voice communication, adapting to the dynamic channel characteristics of construction scenarios.

[0050] 3. This invention uses an edge processing unit to call standard feature vectors of typical construction conditions from a feature library, calculates the similarity between real-time feature vectors and standard vectors to determine candidate conditions and generate dynamic thresholds (including equipment, personnel, and environmental thresholds). Then, based on the comparison between the global fusion results and the dynamic thresholds, it generates hierarchical control strategies (minor parameter adjustments and severe emergency interventions). Combined with the instruction-analysis closed-loop module of the remote monitoring unit (encrypted instruction issuance, blockchain evidence storage, and historical data trend analysis), it effectively solves the problems of limited control strategy coverage dimensions and lack of hierarchical response capabilities that match construction conditions. This achieves dynamic adaptation and control of multi-dimensional construction conditions and closed-loop monitoring of "instruction issuance-result feedback". Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the system framework of the present invention;

[0052] The meanings of the labels in the diagram are as follows:

[0053] 100. On-site sensing unit; 110. Personnel positioning module; 120. Equipment sensing module; 130. Environmental monitoring module;

[0054] 200. Edge processing unit; 210. Data preprocessing module; 220. Feature extraction module; 230. Fusion decision module;

[0055] 300. 5G communication unit; 310. Communication interface module; 320. Link monitoring module; 330. Scene perception module; 340. Resource adaptation module; 350. Anti-interference processing module; 360. Voice-data collaboration module;

[0056] 400. Remote monitoring unit; 410. Data management module; 420. Visual interaction module; 430. Command-analysis closed-loop module. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0058] like Figure 1 As shown, this embodiment provides a remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing, including:

[0059] The field sensing unit 100 uses an industrial-grade sensor array to collect data on the location of construction personnel, equipment operating parameters, and environmental physical parameters, and generates a monitoring data stream with timestamps.

[0060] In this embodiment, the industrial-grade sensor array of the field sensing unit 100 includes a personnel positioning module 110, an equipment sensing module 120, and an environmental monitoring module 130, wherein:

[0061] The personnel positioning module 110 uses a Beidou positioning sensor, which is integrated inside the smart safety helmet worn by construction workers. The sensor is connected to the built-in microcontroller of the safety helmet through an industrial standard serial interface. It collects the three-dimensional position coordinates of construction workers at a frequency that meets the safety tracking needs of construction workers. During the collection process, the power consumption of the sensor meets the industrial low-power equipment operation standards.

[0062] Specifically, when integrating the BeiDou positioning sensor into the smart safety helmet, a module adapted to the helmet's wearing comfort is selected and fixed via pre-reserved mounting slots and buckles in the outer shell and inner lining of the helmet to prevent displacement during operation. The sensor uses an RS-485 industrial standard serial interface to connect with the helmet's built-in microcontroller, adapting to the transmission distance requirements of the construction site to reduce signal attenuation. The microcontroller has a local storage unit to temporarily store the collected three-dimensional position coordinates of personnel, preventing data loss due to momentary communication interruptions. The acquisition frequency is dynamically adjusted according to the scenario: once every 3 seconds in open outdoor areas (such as the main structure pouring area), and once every 1 second in enclosed indoor or equipment-dense areas (such as underground garage decoration areas or scaffolding erection areas). It can be modified in real time via commands issued by the remote monitoring unit 400, and the microcontroller executes the switching. Sensor power consumption is controlled by the microcontroller's sleep mechanism. When the personnel's position remains unchanged for 5 consecutive minutes, the sensor enters sleep mode while retaining intermittent wake-up acquisition function. After detecting a position change, it immediately resumes normal mode to balance acquisition needs and power consumption.

[0063] The equipment sensing module 120 includes a vibration sensor and a temperature sensor. The vibration sensor is fixed to the surface of the load-bearing component of the key construction equipment (such as the tower crane motor, the construction elevator drive shaft, etc.), and the temperature sensor is attached to the heat-generating component inside the equipment distribution box. Both the vibration sensor and the temperature sensor collect data at a frequency that is adapted to the monitoring requirements of the equipment's operating status. The output signals of the vibration sensor and the temperature sensor are transmitted after analog-to-digital conversion.

[0064] Specifically, vibration sensors are fixed according to the material of the load-bearing components of key construction equipment. Cast iron components (such as tower crane motors and construction elevator drive bearing surfaces) are fixed with stainless steel bolts that match the reserved mounting holes to ensure no gaps between the sensor and the component surface. Aluminum alloy components are fixed with a strong magnetic base, and a thin silicone pad is pasted between the base and the component to prevent scratches. Before attaching temperature sensors to heat-generating components (such as contactors and circuit breaker terminals) in the equipment's distribution box, the surface of the components is cleaned of dust and oil, thermally conductive silicone is applied, and high-temperature resistant tape is used for auxiliary fixing. The lead wires are routed along the inner wall of the distribution box and pass through... The wire clips are fixed to prevent contact with exposed terminals; the vibration sensor and temperature sensor complete the analog-to-digital conversion through the built-in conversion module. After conversion, the digital signal is transmitted to the local acquisition terminal of the equipment via the SPI interface, and then the acquisition terminal is connected to the main controller of the field sensing unit 100 through the industrial bus; the acquisition frequency is adapted to the operating characteristics of the equipment. The vibration sensor of high-speed operating equipment (such as tower crane motor) is set to once every 0.5 seconds, the vibration sensor of low-speed operating equipment (such as construction mixer) is set to once every 2 seconds, and the temperature sensor is uniformly set to once every 1 second to adapt to the temperature change rate of the heat-generating components.

[0065] The environmental monitoring module 130 uses dust sensors, noise sensors, and temperature and humidity sensors, which are distributed according to the spacing of the key construction areas (material storage area, scaffolding perimeter). The dust sensors, noise sensors, and temperature and humidity sensors are connected to the main controller of the field sensing unit 100 through an industrial bus, and collect data at a frequency that adapts to the monitoring needs of environmental parameter changes. The sensor housing is designed with a protection level that adapts to construction dust and humid environments.

[0066] Specifically, dust, noise, and temperature / humidity sensors are distributed according to the characteristics of key construction areas. In material storage areas (such as steel bar yards and sand and gravel yards), one set is placed every 20-30 meters, and around scaffolding, one set is placed every 15-20 meters. Sensor installation locations are adapted to environmental requirements. Dust and noise sensors are installed on metal supports 1.5-2 meters high, fixed to walls, temporary guardrails, or other fixed structures. Temperature and humidity sensors are installed 1 meter above the ground, avoiding direct sunlight and rain (such as inside temporary protective sheds). The sensors are connected to the 100-level field sensing unit main controller via a Modbus industrial bus, adapting to multi-sensor networking requirements. The sensor housing is made of ABS engineering plastic, with rubber sealing rings at the interfaces to withstand dusty and humid environments and protect internal components for normal operation.

[0067] In this embodiment, the field sensing unit 100 generates a timestamped monitoring data stream, specifically including:

[0068] With the built-in verification function of the industrial-grade sensor array, when collecting the location of construction personnel, equipment operating parameters and environmental physical parameters, the power supply stability and signal transmission integrity of each sensor are monitored simultaneously. When the power supply of the sensor exceeds its normal operating range or when packet loss occurs in the signal transmission, an abnormality mark is automatically marked next to the timestamp of the corresponding collected data.

[0069] By binding the time stamp generation moment to the actual sensor acquisition action, the time stamp is generated synchronously at the moment the sensor triggers acquisition, avoiding time deviation caused by data temporary storage and ensuring the consistency of the time stamp and parameter acquisition behavior in the time dimension.

[0070] The generated monitoring data stream is distinguished and marked according to parameter type (construction personnel location, equipment operating parameters, environmental physical parameters). The marking information is embedded in the data stream header, so that the edge processing unit 200 can directly identify the data category and match the corresponding processing strategy.

[0071] Specifically, the main controller implements a built-in verification function. When receiving sensor data, the main controller simultaneously reads the power supply voltage signal and the transmission data packet. If the power supply voltage exceeds the normal operating range of the sensor or the data packet has missing fields, it automatically adds a "power supply abnormality" or "transmission packet loss" label next to the corresponding data timestamp. The binding of the timestamp and the acquisition action is achieved through the main controller's trigger command. When the main controller sends an acquisition trigger command to the sensor, it simultaneously starts the internal clock to generate a timestamp. After receiving the command, the sensor immediately acquires the data and directly associates it with the timestamp, avoiding time deviations caused by temporary data storage. The data stream is distinguished and marked according to parameter type. The main controller embeds an identification field in the header of the data stream: "P-001" corresponds to "personnel location data", "E-002" corresponds to "equipment operating parameters", and "ENV-003" corresponds to "environmental physical parameters", which facilitates the edge processing unit 200 to directly identify and match processing strategies.

[0072] Meanwhile, the main controller categorizes and temporarily stores data streams in different local cache areas according to parameter type, and uploads them with priority according to "personnel data-equipment data-environment data" when communicating with the edge processing unit 200, ensuring that critical data is transmitted first.

[0073] Edge processing unit 200 is communicatively connected to field sensing unit 100. It builds a multi-source data processing engine based on edge computing architecture, and achieves accurate integration of heterogeneous data through an improved adaptive weighted fusion algorithm. It also generates real-time control strategies by combining the dynamic working condition model.

[0074] Understandably, the edge processing unit 200's hardware deployment specifically includes: using an industrial-grade edge gateway as the core hardware carrier. This gateway has multiple interface expansion capabilities such as RS-485, Ethernet, and 5G NR, and can directly establish stable communication with the field sensing unit 100 and the 5G communication unit 300. The gateway shell adopts a protective design adapted to the dusty and humid environment of the construction site. The multi-source data processing engine is implemented by deploying lightweight data processing software in the edge gateway. The software supports modular calls and can independently start data preprocessing, feature extraction, or fusion decision-making functions according to the needs of the construction scenario, so as to allocate computing resources on demand. At the same time, the gateway is configured with a local database to store the working condition feature library, historical fusion data, and preprocessed intermediate data, ensuring the continuity of data processing and decision generation.

[0075] In this embodiment, the edge processing unit 200 includes a data preprocessing module 210, a feature extraction module 220, and a fusion decision module 230, wherein:

[0076] The data preprocessing module 210 receives the time-stamped monitoring data stream output by the field sensing unit 100, performs time-domain smoothing on the data through a sliding window to filter out high-frequency interference signals, completes the data marked as abnormal in the data stream (including the abnormal identifiers marked by the field sensing unit 100) using an interpolation algorithm based on the distribution law of data from the same period of similar sensors, and physically truncates extreme data that exceed the sensor's range (the truncation value is set based on the sensor's rated working range). During the processing, the one-to-one correspondence between the data and the original timestamp is maintained.

[0077] Specifically, when receiving the timestamped monitoring data stream output by the field sensing unit 100, the data is first classified and received according to the parameter type identifier in the data stream header (e.g., "P-001" represents personnel location data, "E-002" represents equipment operating parameters, and "ENV-003" represents environmental physical parameters). When performing sliding window time-domain smoothing processing, the window size is adapted according to the data acquisition frequency. For example, for equipment temperature data acquired at a frequency of 1 time / second, the window size is set to 5 sampling points. High-frequency interference is filtered out by calculating the average value of the data within the window, using a formula referencing the sliding average logic. Interpolation is used to complete the data marked as abnormal. Personnel location data uses linear interpolation (based on normal data from three sampling periods before and after abnormal data), while environmental parameter data uses nearest neighbor interpolation (based on data from the same type of sensor in the same area during the same period). For extreme data exceeding the sensor's range, a cutoff value is set according to the sensor's rated operating range (e.g., if a temperature sensor's range is -20℃ to 80℃, data below -20℃ or above 80℃ will be cut off to -20℃ and 80℃ respectively). All preprocessing operations maintain a one-to-one correspondence between the data and the original timestamp. The preprocessed data is temporarily stored in the edge gateway's local cache, awaiting call from the feature extraction module 220.

[0078] The feature extraction module 220 performs dimensional feature extraction on the preprocessed data. It extracts the direction vector of the movement trajectory and the regional stay feature from the construction personnel location data, extracts the steady-state component and transient fluctuation feature of the operating state from the equipment operating parameters, and extracts the gradient feature and cumulative effect feature of parameter changes from the environmental physical parameters. All extracted features are output in the form of standardized numerical vectors, and the vector dimension corresponds one-to-one with the data type.

[0079] Specifically, when extracting features from the preprocessed construction worker location data, the three-dimensional location coordinates of two adjacent timestamps are used (assuming the coordinates of the previous time point are...). The coordinates at the next time step are ), calculate the direction vector of the movement trajectory By determining that the change range of a person's position coordinates within a certain area (not exceeding 1 meter in the X, Y, and Z axes) and the duration exceeds 5 sampling periods, the area dwell characteristics are extracted.

[0080] Specifically, when extracting features from equipment operating parameters, the moving average method with the same window size as in the data preprocessing stage is used to calculate the steady-state component (such as the moving average result of equipment vibration data), and the transient fluctuation features are extracted by calculating the absolute value of the difference between the data at each sampling point and the steady-state component.

[0081] Specifically, when extracting features from environmental physical parameters, gradient features are calculated by the parameter difference between two adjacent time stamps (such as the dust concentration at the later time minus the concentration at the previous time), and cumulative effect features are extracted by the number of consecutive sampling periods where statistical parameters exceed the normal range.

[0082] Furthermore, all extracted features are converted into standardized numerical vectors, with the vector dimensions corresponding one-to-one with the data type (e.g., personnel location data corresponds to a 2D vector, equipment operating parameters correspond to a 2D vector, and environmental parameters correspond to a 2D vector). The vector values ​​are processed through a normalization method of "(actual feature value - historical minimum feature value) / (historical maximum feature value - historical minimum feature value)" to ensure that the numerical range of different types of features is consistent.

[0083] The fusion decision module 230 aligns the multi-dimensional feature vectors output by the feature extraction module 220 with spatial coordinates, assigns fusion coefficients to different features through the feature contribution evaluation model, and integrates the weighted feature vectors into a unified working condition evaluation matrix. The matrix dimension adapts to the multi-source data processing engine calculation requirements of the edge processing unit 200.

[0084] Specifically, when performing spatial coordinate alignment on multi-dimensional feature vectors, a two-dimensional plane coordinate system is established with a fixed corner point of the construction site as the origin (elevation data is retained separately as the third dimension). Personnel location features, equipment location associated operation features, and environmental features of the environmental monitoring area are uniformly mapped to this coordinate system to ensure that different types of features can correspond in spatial location.

[0085] Specifically, when allocating fusion coefficients through the feature contribution evaluation model, the analytic hierarchy process (AHP) is used to determine the weights. For example, in the equipment fault early warning scenario, the fusion coefficient for equipment operating parameters is set to 0.6, and the fusion coefficients for personnel location and environmental parameters are each set to 0.2. In the personnel safety management scenario, the fusion coefficient for personnel location is set to 0.5, the fusion coefficient for environmental parameters is set to 0.3, and the fusion coefficient for equipment operating parameters is set to 0.2. The weighted feature vectors are then integrated into a condition evaluation matrix in the format of "rows representing feature types and columns representing timestamps". The matrix dimensions are adapted to the edge gateway's computing capabilities (e.g., when processing 100 timestamps of data each time, the matrix dimensions are set to 6 rows × 100 columns, with the 6 rows corresponding to 2 features for personnel, 2 features for equipment, and 2 features for the environment), ensuring efficient reading and calculation by the multi-source data processing engine.

[0086] In this embodiment, the edge processing unit 200 achieves accurate integration of heterogeneous data through an improved adaptive weighted fusion algorithm, including the following steps:

[0087] S240.1 Receives multi-source heterogeneous data processed by data preprocessing module 210. The data includes construction personnel location data, equipment operating parameters, and environmental physical parameters. Assigns a unique data source identifier to each type of data and initializes the real-time reliability coefficient of each data source. Initial value; can be used to determine the real-time reliability coefficient. The initial value is set to 0.8.

[0088] S240.2 Calculate the real-time reliability coefficient of each data source based on the data source identifier. By analyzing the temporal consistency (the range of fluctuation in continuous sampled values) and spatial correlation (the degree of deviation of data from similar sensors in the same region), the data can be... Mapped to the [0,1] range, the higher the value, the higher the reliability of the data source;

[0089] Specifically, calculate the real-time reliability coefficient. First, use time sequence consistency score. The formula for assessing the fluctuation of continuous sampled values ​​is as follows: ,in For the first Data source in Time before The coefficient of variation of data in each sampling period ( According to the sampling frequency setting, such as 1 time / second =10), For the first The maximum permissible coefficient of variation for each data source (determined by statistical analysis of historical normal operating data); then, spatial correlation scores are used. The formula for evaluating the data deviation of similar sensors in the same area is as follows: ,in The number of similar sensors in the same area. For the first The first class of data sources The values ​​collected by each sensor For the same region The average value of the data collected by each sensor; finally, calculate... ,in The timing consistency weighting coefficient is 0.6 for device parameters and 0.5 for environmental parameters, and the result is mapped to the interval [0,1].

[0090] S240.3, according to Dynamically allocate fusion weights Satisfying the weight constraint relationship When a certain data source When the reliability falls below a preset threshold, the corresponding data source will be automatically downgraded. The percentage, the released weight share, is determined by other highly reliable data sources according to... Proportional sharing;

[0091] Specifically, fusion weights Allocation satisfies When all ( When the preset reliability threshold is set to 0.5, the allocation is as follows: ;

[0092] When it exists ( When assigning a low-reliability data source number, adjust according to the following formula: ;in, The minimum weight is set to 0.1; Indicates the first Each data source (sensing / transmission nodes, etc., participating in data fusion) at time... The real-time reliability coefficient is used to quantify the reliability of the data source at this moment. Indicates when the first When the reliability of the first data source is insufficient, except for the first In addition to the first data source, the second Data sources at time The real-time reliability coefficient.

[0093] S240.4. Data is processed using a hierarchical fusion method, first through a local fusion operator. Data from similar sensors is fused (e.g., vibration parameter fusion from multiple devices, environmental parameter fusion from multiple regions), and then fused using a global fusion operator. The results of different types of local fusion are integrated to generate a global fusion result. ;

[0094] Specifically, a layered fusion approach is adopted, and local fusion calculates the fusion result of similar data using the following formula: ,in For the first The partial fusion result of the data source class This indicates the number of "sub-data sources (or acquisition nodes)" participating in local fusion; global fusion calculates the global fusion result using the following formula. : .

[0095] S240.5, Through deviation verification factor verify The effectiveness, Based on the statistical patterns of deviations between historical fusion data and actual working conditions, if... The deviation exceeds If the range is not cleared, return to S240.3 to readjust the weight allocation until a valid fusion result is generated.

[0096] Specifically, through ,verify Validity, among which This refers to actual operating conditions (such as manual inspection records). This is the deviation check factor (determined by the 95th percentile of historical data). If it exceeds the range, return to S240.3 to readjust the weights.

[0097] In this embodiment, the edge processing unit 200 generates a real-time control strategy by combining the dynamic model of the operating conditions, including the following steps:

[0098] S250.1, Recall the data stored in the operating condition feature library. Typical construction condition standard feature vector Standard feature vector Generated by clustering historical normal operating condition data, each standard feature vector contains equipment operation feature components, personnel distribution feature components, and environmental parameter feature components;

[0099] Specifically, the operating condition feature library is stored in the local database of the edge gateway, and the standard feature vectors are... ( The number of standard working condition categories is set according to the main construction procedures, such as... The standard feature vector is generated by K-means clustering from the historical normal operating data of the three months prior to system deployment. Each standard feature vector contains three types of feature components: equipment operation, personnel distribution, and environmental parameters.

[0100] S250.2 Extract the global fusion result after processing with the improved adaptive weighted fusion algorithm. Corresponding real-time feature vector Through the vector space distance function calculate With each standard eigenvector similarity distance Select The three smallest standard operating conditions are selected as the current matching candidate set;

[0101] Specifically, using the Euclidean distance formula Calculate real-time feature vectors Compared with standard feature vectors similarity distance ,in For vector dimensions, such as =6; Select The three smallest standard operating conditions are selected as the candidate set;

[0102] S250.3. Based on the distribution patterns of historical abnormal data of candidate operating conditions, generate dynamic thresholds for the current operating conditions. , The correction amount is positively correlated with the historical anomaly deviation of the candidate set operating conditions, and the corrected amount is... This includes equipment safe operation threshold ranges, personnel safe density thresholds, and environmental parameter early warning thresholds;

[0103] Specifically, dynamic threshold Corrected using the following formula: ,in The baseline threshold for candidate operating conditions (determined by historical normal operating conditions). This is a correction factor (set to 0.1). The historical average deviation of the candidate operating condition is calculated using the following formula: , This refers to the number of historical anomalies. For the first (Global fusion result at the time of sub-anomaly), corrected It includes three sub-thresholds: equipment, personnel, and environment;

[0104] S250.4, Combine the global fusion results With dynamic threshold When a comparison is performed, When a certain parameter exceeds the corresponding threshold range, a graded control strategy is generated based on the degree of deviation: when the deviation is slight, a parameter adjustment instruction is generated (such as fine-tuning of equipment operating frequency or starting of environmental control equipment); when the deviation is severe, an emergency intervention instruction is generated (such as equipment shutdown instruction or personnel evacuation warning), and the control strategy is encapsulated and transmitted to the 5G communication unit 300.

[0105] Specifically, through Classify the degree of deviation; among them, The coefficient for slight deviation is set to 0.1; The severe deviation coefficient is set to 0.2. Meanwhile, mild deviations generate parameter adjustment instructions (such as fine-tuning of equipment operating frequency), and severe deviations generate emergency intervention instructions (such as equipment shutdown and personnel evacuation warnings). All instructions are encapsulated in a fixed format (including instruction type, target device ID, and execution time limit) and a CRC checksum is added before being transmitted to the 5G communication unit 300.

[0106] The 5G communication unit 300 is connected to the edge processing unit 200 and the remote monitoring unit 400 respectively. Through the "scene perception-resource adaptation" dual closed-loop transmission mechanism, combined with predictive slicing scheduling and multimodal anti-interference protocol, it realizes the ultra-low latency transmission of high-priority data and seamless coordination of voice communication, adapting to the dynamic channel characteristics of the construction scenario.

[0107] Specifically, the 5G communication unit 300 employs a combined architecture of an industrial-grade 5G communication module and an industrial router in its hardware deployment. The core module is the TurboXT75 series cellular communication module, which supports 5G NR Sub-6SA / NSA dual-mode and is backward compatible with 4G / 3G networks. It integrates with the CM520-6XX series industrial router via a PCIe interface. The industrial router is equipped with four Gigabit Ethernet LAN ports and one Gigabit WAN port. The WAN port establishes a wired connection with the edge gateway of the edge processing unit 200 via an Ethernet link, ensuring stable data transmission. The device features a metal casing design, providing dust and moisture protection, making it suitable for harsh environments with high dust and humidity, such as construction sites. It also supports dual SIM card redundancy backup, automatically switching to the backup card in case of primary SIM link failure. The router has a built-in GPS module and multiple SMA antenna interfaces, connecting to the 5G primary and secondary antennas and the WIFI antenna respectively. The antennas are deployed at high points on the construction site (such as the top of a tower crane or a temporary signal tower), and the antenna azimuth angle is adjusted to reduce signal attenuation caused by building obstruction.

[0108] Specifically, the data transmission process of the 5G communication unit 300 includes receiving data output from the edge processing unit 200, first uniformly encapsulating different types of data such as control policy commands and operating condition evaluation matrices. The encapsulation format adopts the structure of "data type identifier + timestamp + payload + CRC checksum". The data type identifier follows the unified system rules (e.g., "CMD-001" represents control commands, "DATA-002" represents monitoring data), maintaining consistency with the output format of the edge processing unit 200. Data encryption uses the AES symmetric encryption algorithm. The encryption key is configured through a pre-shared method and written offline to the secure storage area of ​​the 5G communication unit during system initialization. Key management follows the "one device, one key" principle, using the keyconfig-key command to set the master key and encrypt all transmission keys to prevent key leakage. The encrypted data packets establish a secure transmission channel through the router's VPN tunnel, supporting both IPSecVPN and OpenVPN tunnel modes, dynamically selected according to the access requirements of the remote monitoring center.

[0109] Specifically, the data priority scheduling mechanism of the 5G communication unit 300 includes: adopting a service type-based priority marking strategy to divide data into three levels—emergency control instructions (such as equipment shutdown instructions) are marked as the highest priority (P1), real-time operating data (such as personnel safety alarms) are marked as medium priority (P2), and historical statistical data (such as daily report data) are marked as low priority (P3). The scheduling algorithm adopts a priority scheduling strategy, with the router maintaining three buffer queues corresponding to the priorities. When data arrives, it enters the corresponding queue according to the marking, and the scheduler always extracts data transmission from the highest priority queue. When network congestion occurs, differentiated transmission is achieved by dynamically adjusting the queue weights. The resource allocation ratio of queues P1, P2, and P3 is 6:3:1 to ensure that emergency instructions can still be transmitted with priority even when the network load is high. When the data frame length exceeds the MTU (Maximum Transmission Unit), a fragmentation and reassembly mechanism is adopted. Each fragment carries a fragment offset and a total fragment count identifier, and reassembly is completed at the receiving end.

[0110] In this embodiment, the 5G communication unit 300 includes a communication interface module 310 and a link monitoring module 320, wherein:

[0111] The communication interface module 310 establishes connections with the edge processing unit 200 and the remote monitoring unit 400 respectively. It connects with the edge processing unit 200 via an industrial Ethernet interface (compliant with the IEEE 802.3 standard) and transmits data through shielded twisted-pair cable. It also integrates NTP time synchronization function to keep the clock reference of the edge processing unit 200 consistent, ensuring that the time dimension of the timestamped monitoring data stream, global fusion results and parameter adjustment instructions is unbiased during transmission. It connects with the remote monitoring unit 400 via a 5G NR wireless interface (compliant with 3G PPR15 and above standards), supports independent networking mode, and the interface transmission capability adapts to the bandwidth requirements of the "scene perception-resource adaptation" dual closed-loop transmission mechanism, providing a basic transmission link for high-priority data and voice communication.

[0112] The link monitoring module 320 collects the link status parameters of the communication interface module 310 in real time, including data transmission rate, bit error rate and signal strength. When the link bit error rate exceeds the transmission reliability threshold of the adapted construction scenario, or the signal strength is lower than the critical value to ensure continuous data transmission, a link abnormality signal is generated and fed back to the "scenario perception-resource adaptation" dual closed-loop transmission mechanism. This provides the link status basis for resource adaptation adjustment, ensures dynamic matching between link status and resource allocation, and adapts to the dynamic channel characteristics of the construction scenario.

[0113] In this embodiment, the 5G communication unit 300 further includes a scene perception module 330 and a resource adaptation module 340, wherein:

[0114] The scene perception module 330 receives the working condition feature vector output by the edge processing unit 200 in real time, and simultaneously collects the 5G channel parameters of the construction area. It then integrates the working condition feature vector and the channel parameters by timestamp to generate a scene perception matrix. ,and The time dimension and the global fusion result of the edge processing unit 200 Maintaining synchronization provides spatiotemporally consistent scenario data support for resource adaptation; among which As a dimension of working condition characteristics, For channel parameters;

[0115] Specifically, constructing a scene-aware matrix First, calculate the characteristic vector of the working condition. Three core characteristics were selected: construction intensity, equipment operating load, and personnel density. The construction intensity was calculated using the following formula. : ;in, This is the equipment load weighting coefficient. The maximum allowable personnel density in the construction area; thus obtaining ;

[0116] Then calculate the channel parameter vector By selecting three core parameters—signal received power, Doppler frequency offset, and channel bandwidth—we obtained... ;

[0117] Finally, associate by timestamp. and Construct a 3×3 scene perception matrix Updated every 100ms This ensures that resource adaptation is based on real-time scenario data.

[0118] Resource adaptation module 340 based on scene-aware matrix Perform predictive slice scheduling when medium construction intensity At that time, a dedicated network slice is allocated to high-priority data such as equipment operating parameters. ,in The threshold for determining high-intensity work; when Medium signal blocking coefficient At the same time, dynamically expand the slice bandwidth and adapt to the modulation and coding scheme. ( With Doppler frequency shift (Adjusting the order of changes) ensures that network slice resources are matched with scenario requirements in real time, enabling ultra-low latency transmission of high-priority data (ensuring latency). Not exceeding the construction safety response delay threshold );in This is the critical value for signal attenuation.

[0119] Specifically, when performing predictive slice scheduling, the signal occlusion coefficient is first calculated. The formula is: ;in, For unobstructed reference signal received power, This is the minimum signal power that the device can receive;

[0120] when The signal is determined to be blocked at that time; among which This is the critical value for signal attenuation, based on a preset signal attenuation test at the construction site. For example, if... Simultaneously calculate construction strength. ;in This is the equipment load weighting coefficient. The maximum allowable personnel density in the construction area; when At that time, a dedicated network slice is allocated to high-priority data such as equipment operating parameters. ;in The threshold for determining high-intensity work is preset based on historical statistics of high-intensity construction conditions, such as taking... ;

[0121] like Through formula Dynamically expand the slice bandwidth; among which Based on the basic slice bandwidth, This is the bandwidth expansion factor;

[0122] Simultaneously, based on Doppler frequency shift Adjusting the modulation and coding scheme Order, when Different intervals correspond to different S-order;

[0123] Finally, the formula is used. Verify high-priority data latency Does it meet the requirements? If not, further adjust the bandwidth or MCS order; among which... For data frame length, For slice transmission rate, This is to fix the processing delay.

[0124] In this embodiment, the 5G communication unit 300 further includes an anti-interference processing module 350 and a voice-data coordination module 360, wherein:

[0125] The anti-interference processing module 350 operates based on a multi-mode anti-interference protocol, and its built-in channel monitoring component collects the electromagnetic interference intensity of the construction area in real time. With multipath effect parameters When sudden electromagnetic interference is detected ( Reaching the interference tolerance threshold When, based on the detected interference frequency band Generate frequency hopping pattern and set frequency hopping interval. When signal obstruction in a fixed area is detected ( Reaching the signal attenuation threshold When ), 3D beamforming technology is activated, through angle Adjust signal propagation direction (related to the height of obstructions) With transmission distance (Diffraction of obstructions to ensure low bit error rate in signal transmission) Not exceeding the industrial-grade transmission reliability threshold ;

[0126] Specifically, through the formula Calculate electromagnetic interference intensity ;in To interfere with signal power, For the useful signal power, when Frequency hopping is initiated when the threshold is not lowered.

[0127] Through formula Calculate the frequency hopping interval ;in For interference frequency band bandwidth, To protect bandwidth, frequency hopping patterns are generated to avoid interfering frequency bands;

[0128] Through formula Calculate the elevation angle of 3D beamforming ,in For the height of the obstruction, For 5G antenna height, To determine the transmission distance, adjust the beam to bypass obstructions; then use the formula... Verify transmission error rate If the requirements are not met, the frequency hopping interval or beam angle should be readjusted. for function, This refers to the signal-to-noise ratio.

[0129] The 360 ​​Voice-Data Collaboration Module utilizes 5G VoNR technology to achieve seamless collaboration between voice communication and high-priority data, mapping voice signals to QoS streams. (Guarantee delay) Mapping high-priority data to QoS streams (Guaranteed bit rate) ); adopts a transmission frame structure of "voice frame embedded data subframe", each A high-priority data subframe is embedded after each consecutive voice frame, and the timestamp of the data subframe is aligned with that of the voice frame to ensure uninterrupted voice communication and zero data transmission delay, achieving collaborative adaptation between the two. This is the frame synchronization coefficient, adapted to the speech sampling rate.

[0130] Specifically, through the formula Calculate the frame synchronization coefficient Take the integer part Ensure each Each audio frame is embedded in one data subframe, and the timestamps of the data subframes are aligned with those of the audio frames to ensure uninterrupted audio playback and zero data latency; For data subframe transmission time, For the transmission time of a single voice frame, The length of the data subframe. For data subframe transmission rate, This refers to the speech sampling rate.

[0131] It is understood that the link maintenance and reliability assurance mechanism of the 5G communication unit 300 in this embodiment specifically includes: real-time monitoring of communication link quality, evaluating link status by periodically collecting reference signal received power (SSB-RSRP) and signal-to-interference-plus-noise ratio (SSB-SINR), and initiating a link optimization process when SSB-RSRP remains below -105dBm for five consecutive sampling periods. The optimization measures are: adjusting beam direction and digital tilt angle through the router's built-in ACP (Automatic Cell Optimization) function; if the link quality still does not improve after optimization, a network handover mechanism is triggered, automatically switching from 5G NR to the 4G LTE network, and sending a link handover alarm to the edge processing unit 200. Data transmission adopts the Selective Repeat ARQ (SR-ARQ) protocol. The sending end sets an independent timer for each data frame; if no acknowledgment (ACK) is received within the timeout period, only the unacknowledged frame is retransmitted. The receiving end buffers out-of-order but correct frames, and completes in-order delivery after the missing frames are retransmitted. The system maintains a long-term connection with the remote monitoring center by sending heartbeat packets periodically (every 30 seconds). The heartbeat packet contains the device ID and the current link status information. If no heartbeat response is received for 3 consecutive times, the router will be soft-rebooted to restore the connection.

[0132] It should be added that the status feedback and management function of the 5G communication unit 300 in this embodiment specifically includes real-time recording of the transmission status during data transmission, including indicators such as the number of successfully transmitted frames, the number of retransmissions, and the packet loss rate. A link quality report is generated every 5 minutes and uploaded to the edge processing unit 200. Remote management is supported; maintenance personnel can access the router configuration interface via HTTPS or SSH to adjust parameters, upgrade firmware, and query logs. Log information includes access authentication records, link switching events, encryption failure alarms, etc., stored in SYSLOG format and supported for power-off retention. When an abnormal access attempt is detected (such as three consecutive incorrect password attempts), firewall rules are automatically triggered to temporarily block the IP address and record the intrusion log, ensuring device access security.

[0133] The remote monitoring unit 400 receives edge processing results through the 5G communication unit 300, stores historical data and displays real-time status, and sends parameter adjustment instructions to the edge processing unit 200 to form a closed-loop monitoring link.

[0134] In this embodiment, the remote monitoring unit 400 includes a data management module 410, a visualization interaction module 420, and a command-analysis closed-loop module 430, wherein:

[0135] The data management module 410 uses a hierarchical storage architecture to process edge processing results and real-time global fusion results. eigenvectors Millisecond-level read and write operations are performed using an in-memory database to ensure real-time data access efficiency; historical operating data is stored in a columnar database partitioned by time dimension, supporting fast retrieval of TB-level data, and the storage strategy is dynamically adapted to the transmission bandwidth of the 5G communication unit 300.

[0136] Specifically, the layered storage architecture for edge processing results includes: real-time global fusion results and feature vectors are processed using a Redis in-memory database for millisecond-level read and write operations. The Redis cluster adopts a master-slave replication mode, with the master node responsible for data writing and the slave nodes handling read requests, ensuring that real-time data (such as equipment status updates and personnel location changes) is read and written within 10ms. Historical operating data (such as daily equipment operation logs and historical environmental parameter curves) is processed using a ClickHouse columnar database, partitioned and stored according to the "year-month-day" time dimension, supporting fast retrieval of TB-level data (such as querying the vibration data of a tower crane in the past 3 months with a response time of no more than 1 second).

[0137] Meanwhile, the storage strategy is dynamically adapted to the transmission bandwidth of the 5G communication unit 300. That is, the data management module 410 receives the link quality report uploaded by the 5G communication unit 300 in real time. When the report shows that the transmission bandwidth is less than 10Mbps, the historical data archiving frequency is automatically adjusted from once every 5 minutes to once every 15 minutes to prioritize real-time data transmission. When the bandwidth returns to normal, the original archiving frequency is restored.

[0138] The visualization interaction module 420 is used to build a digital twin visualization interface for the construction scene, which transforms the edge processing results into dynamic identification of equipment status, dynamic distribution trajectory of personnel and real-time curves of environmental parameters. It supports synchronous refresh of the interface of multiple terminals, and the refresh frequency is consistent with the global fusion cycle of the edge processing unit 200.

[0139] Specifically, based on the CAD drawings and BIM models of the construction project, a 1:1 scale digital twin scene is built in the Unity3D engine to digitally map the building structure, equipment distribution, temporary facilities, etc. of the construction site. The equipment operation parameters output by the edge processing unit 200 are converted into dynamic status indicators of the equipment in the digital twin interface (e.g., when the tower crane motor temperature is too high, the corresponding part of the tower crane model in the digital twin displays a red warning; when the equipment is running normally, it displays green). The dynamic distribution trajectory of personnel is rendered by generating a dynamic point sequence with timestamps in the digital twin scene, and the color of the points distinguishes different types of work (e.g., electricians are yellow and welders are blue). The real-time curves of environmental parameters are drawn using the ECharts component and displayed in the sidebar of the interface as a line chart, supporting zooming of the time range (e.g., viewing the dust concentration changes in the past hour or the past day).

[0140] Furthermore, the interface synchronization refresh of multiple terminals is implemented based on the WebSocket protocol. The server and the client (such as the monitoring center screen, the mobile APP of the maintenance personnel) establish a long connection, and the refresh frequency is consistent with the global fusion cycle of the edge processing unit 200 (such as once every 100ms) to ensure that the interface status of each terminal is synchronized in real time.

[0141] The instruction-analysis closed-loop module 430 realizes the encrypted issuance of parameter adjustment instructions and historical data-driven analysis: the instructions issued to the edge processing unit 200 are encrypted using the national cryptographic SM4 algorithm and digitally signed, and the entire instruction process is recorded through blockchain notarization technology; at the same time, construction trend analysis is performed based on historical data and analysis results are generated to provide data support for instruction generation, forming a closed-loop link of "instruction issuance - result feedback - trend analysis - instruction optimization".

[0142] Specifically, parameter adjustment commands (such as device operating frequency adjustment commands and sensor acquisition frequency adjustment commands) issued to the edge processing unit 200 are encrypted using the national cryptographic SM4 symmetric encryption algorithm. The encryption key is generated and stored through a hardware encryption module. After the command is generated, it is first encrypted using the SM4 algorithm, and then a digital signature generated by the RSA algorithm (containing information such as command generation time and generating terminal ID) is attached. The entire command process is recorded using blockchain evidence storage technology. The blockchain nodes are composed of the monitoring center server and the edge server where the edge processing unit 200 is located. The encrypted content of the command, the issuance time, the receipt confirmation time, and other information are stored on the chain in the form of blocks to ensure that the command is traceable and tamper-proof.

[0143] Simultaneously, construction trend analysis is conducted based on historical data. Time series analysis methods (such as ARIMA models) are used to model historical working condition data (such as equipment vibration data and environmental temperature and humidity data) to predict the construction status trend in the future (such as predicting whether the dust concentration in a construction area will exceed the standard in 2 hours) and generate analysis result reports. When the trend analysis shows potential risks (such as predicting that the equipment vibration value will exceed the safety threshold), the instruction generation logic is automatically triggered to provide data support for the closed-loop link of "instruction issuance - edge processing result feedback - historical data trend analysis - instruction optimization issuance".

[0144] Those skilled in the art will understand that the process of implementing all or part of the steps of the above embodiments can be carried out by hardware or by a program instructing the relevant hardware.

[0145] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing, characterized in that, include: The field sensing unit (100) adopts an industrial-grade sensor array to collect the location of construction personnel, equipment operating parameters and environmental physical parameters, and generate a monitoring data stream with timestamps; An edge processing unit (200) is communicatively connected to a field sensing unit (100). It constructs a multi-source data processing engine based on an edge computing architecture, achieves accurate integration of heterogeneous data through an improved adaptive weighted fusion algorithm, and generates real-time control strategies by combining a dynamic working condition model. The edge processing unit (200) includes a data preprocessing module (210). The data preprocessing module (210) receives the time-stamped monitoring data stream output by the field sensing unit (100), performs time-domain smoothing processing on the data through a sliding window to filter out high-frequency interference signals, completes the data marked as abnormal in the data stream using an interpolation algorithm based on the distribution law of data from the same type of sensor at the same time, and physically truncates the extreme value data that exceeds the sensor range. During the processing, the one-to-one correspondence between the data and the original timestamp is maintained. The edge processing unit (200) achieves accurate integration of heterogeneous data through an improved adaptive weighted fusion algorithm, including the following steps: S240.1 Receive multi-source heterogeneous data processed by the data preprocessing module (210), the data including construction personnel location data, equipment operating parameters and environmental physical parameters, assign a unique data source identifier to each type of data, and initialize the real-time reliability coefficient of each data source. Initial value; S240.2 Calculate the real-time reliability coefficient of each data source based on the data source identifier. By analyzing the temporal consistency and spatial correlation of the data, Mapped to the [0,1] range, the higher the value, the higher the reliability of the data source; S240.3, according to Dynamically allocate fusion weights Satisfying the weight constraint relationship When a certain data source When the reliability falls below a preset threshold, the corresponding data source will be automatically downgraded. The percentage, the released weight share, is determined by other highly reliable data sources according to... Proportional sharing; S240.

4. Data is processed using a hierarchical fusion method, first through a local fusion operator. Data from similar sensors are fused, and then processed using a global fusion operator. The results of different types of local fusion are integrated to generate a global fusion result. ; S240.5, Through deviation verification factor verify The validity, if The deviation exceeds If the range is not cleared, return to S240.3 to readjust the weight allocation until a valid fusion result is generated; The edge processing unit (200) generates a real-time control strategy by combining the dynamic model of the operating conditions, including the following steps: S250.1, Recall the data stored in the operating condition feature library. Typical construction condition standard feature vector The standard feature vector Generated by clustering historical normal operating condition data, each standard feature vector contains equipment operation feature components, personnel distribution feature components, and environmental parameter feature components; S250.2 Extract the global fusion result after processing with the improved adaptive weighted fusion algorithm. Corresponding real-time feature vector Through the vector space distance function calculate With each standard eigenvector similarity distance Select The three smallest standard operating conditions are selected as the current matching candidate set; S250.

3. Based on the distribution patterns of historical abnormal data of candidate operating conditions, generate dynamic thresholds for the current operating conditions. , The correction amount is positively correlated with the historical anomaly deviation of the candidate set operating conditions, and the corrected amount is... This includes equipment safe operation threshold ranges, personnel safe density thresholds, and environmental parameter early warning thresholds; S250.4, Combine the global fusion results With dynamic threshold When a comparison is performed, When a certain parameter exceeds the corresponding threshold range, a graded control strategy is generated based on the degree of deviation: when the deviation is slight, a parameter adjustment instruction is generated; when the deviation is severe, an emergency intervention instruction is generated, and the control strategy is encapsulated and transmitted to the 5G communication unit (300). The 5G communication unit (300) is connected to the edge processing unit (200) and the remote monitoring unit (400) respectively. Through the "scene perception-resource adaptation" dual closed-loop transmission mechanism, combined with predictive slice scheduling and multimodal anti-interference protocol, it realizes the seamless coordination of ultra-low latency transmission of high priority data and voice communication, and adapts to the dynamic channel characteristics of the construction scenario. The 5G communication unit (300) includes a scene perception module (330) and a resource adaptation module (340), wherein: The scene perception module (330) receives the working condition feature vector output by the edge processing unit (200) in real time, and simultaneously collects the 5G channel parameters of the construction area. It integrates the working condition feature vector and the channel parameters according to the timestamp to generate a scene perception matrix. ,and The time dimension and the global fusion result output by the edge processing unit (200) Maintaining synchronization provides spatiotemporally consistent scenario data support for resource adaptation; among which As a dimension of working condition characteristics, For channel parameters; The resource adaptation module (340) is based on a scene-aware matrix. Perform predictive slice scheduling when medium construction intensity At that time, a dedicated network slice is allocated to high-priority data such as equipment operating parameters. ,in The threshold for determining high-intensity work; when Medium signal blocking coefficient At the same time, dynamically expand the slice bandwidth and adapt to the modulation and coding scheme. This ensures that network slice resources are matched with scenario requirements in real time, enabling ultra-low latency transmission of high-priority data; among which This is the critical value for signal attenuation. The 5G communication unit (300) further includes an anti-interference processing module (350) and a voice-data collaboration module (360), wherein: The anti-interference processing module (350) operates based on a multi-modal anti-interference protocol, and its built-in channel monitoring component collects the electromagnetic interference intensity of the construction area in real time. With multipath effect parameters When sudden electromagnetic interference is detected, based on the detected interference frequency band Generate frequency hopping pattern and set frequency hopping interval. When signal obstruction is detected in a fixed area, 3D beamforming technology is activated to adjust the beam pattern by angle. Adjusting the signal propagation direction, diffracting obstructions, and ensuring a low bit error rate in signal transmission. Not exceeding the industrial-grade transmission reliability threshold ; The voice-data collaboration module (360) achieves seamless collaboration between voice communication and high-priority data through 5G VoNR technology, mapping voice signals to QoS streams. Map high-priority data to QoS streams The transmission frame structure adopts "voice frame embedded data subframe" for each... A high-priority data subframe is embedded after each consecutive voice frame, and the timestamp of the data subframe is aligned with that of the voice frame to ensure uninterrupted voice communication and zero data transmission delay. This is the frame synchronization coefficient, adapted to the speech sampling rate; The remote monitoring unit (400) receives edge processing results through the 5G communication unit (300), stores historical data and displays real-time status, and sends parameter adjustment instructions to the edge processing unit (200) to form a closed-loop monitoring link.

2. The remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing as described in claim 1, characterized in that, The industrial-grade sensor array of the field sensing unit (100) includes a personnel positioning module (110), an equipment sensing module (120), and an environmental monitoring module (130), wherein: The personnel positioning module (110) adopts a Beidou positioning sensor, which is integrated inside the smart safety helmet worn by the construction personnel. The sensor is connected to the built-in microcontroller of the safety helmet through an industrial standard serial interface. It collects the three-dimensional position coordinates of the construction personnel at a frequency that meets the safety tracking needs of the construction personnel. During the collection process, the power consumption of the sensor meets the industrial low-power equipment operation standards. The equipment sensing module (120) includes a vibration sensor and a temperature sensor. The vibration sensor is fixed to the surface of the load-bearing component of the key construction equipment, and the temperature sensor is attached to the heat-generating component inside the equipment distribution box. Both the vibration sensor and the temperature sensor collect data at a frequency that is adapted to the monitoring requirements of the equipment's operating status. The output signals of the vibration sensor and the temperature sensor are transmitted after analog-to-digital conversion. The environmental monitoring module (130) uses dust sensors, noise sensors and temperature and humidity sensors, which are distributed according to the spacing of the key construction areas. The dust sensors, noise sensors and temperature and humidity sensors are connected to the main controller of the field sensing unit (100) through an industrial bus. Data is collected at a frequency that adapts to the monitoring needs of environmental parameter changes. The sensor housing is designed with a protection level that adapts to the construction dust and humid environment.

3. The remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing according to claim 2, characterized in that, The field sensing unit (100) generates a timestamped monitoring data stream, specifically including: With the built-in verification function of the industrial-grade sensor array, when collecting the location of construction personnel, equipment operating parameters and environmental physical parameters, the power supply stability and signal transmission integrity of each sensor are monitored simultaneously. When the power supply of the sensor exceeds its normal operating range or when packet loss occurs in the signal transmission, an abnormality mark is automatically marked next to the timestamp of the corresponding collected data. The timestamp generation time is bound to the actual sensor acquisition action. The timestamp is generated synchronously at the moment the sensor triggers acquisition, ensuring the consistency of the timestamp and the parameter acquisition behavior in the time dimension. The generated monitoring data stream is distinguished and marked according to parameter type. The marking information is embedded in the data stream header, so that the edge processing unit (200) can directly identify the data category and match the corresponding processing strategy.

4. The remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing according to claim 1, characterized in that, The edge processing unit (200) further includes a feature extraction module (220) and a fusion decision module (230), wherein: The feature extraction module (220) extracts dimensional features from the preprocessed data. It extracts the direction vector of the movement trajectory and the regional stay feature from the construction personnel location data, extracts the steady-state component and transient fluctuation feature of the operating state from the equipment operating parameters, and extracts the gradient feature and cumulative effect feature of parameter change from the environmental physical parameters. All extracted features are output in the form of standardized numerical vectors, and the vector dimension corresponds one-to-one with the data type. The fusion decision module (230) aligns the multi-dimensional feature vectors output by the feature extraction module (220) with spatial coordinates, assigns fusion coefficients to different features through the feature contribution evaluation model, and integrates the weighted feature vectors into a unified working condition evaluation matrix. The matrix dimension adapts to the multi-source data processing engine calculation requirements of the edge processing unit (200).

5. The remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing according to claim 4, characterized in that, The 5G communication unit (300) includes a communication interface module (310) and a link monitoring module (320), wherein: The communication interface module (310) establishes connections with the edge processing unit (200) and the remote monitoring unit (400) respectively. It connects with the edge processing unit (200) using an industrial Ethernet interface and transmits data through shielded twisted-pair cable. It also integrates NTP time synchronization function to keep the clock reference of the edge processing unit (200) consistent, ensuring that the time dimension of the time-stamped monitoring data stream, global fusion results and parameter adjustment instructions is unbiased during transmission. It connects with the remote monitoring unit (400) using a 5G NR wireless interface, supports independent networking mode, and the interface transmission capability adapts to the bandwidth requirements of the "scene perception-resource adaptation" dual closed-loop transmission mechanism, providing a basic transmission link for high-priority data and voice communication. The link monitoring module (320) collects the link status parameters of the communication interface module (310) in real time, including data transmission rate, bit error rate and signal strength. When the link bit error rate exceeds the transmission reliability threshold of the adapted construction scenario, or the signal strength is lower than the critical value to ensure continuous data transmission, a link abnormal signal is generated and fed back to the "scenario perception-resource adaptation" dual closed-loop transmission mechanism to provide the link status basis for resource adaptation adjustment, ensure the dynamic matching of link status and resource allocation, and adapt to the dynamic channel characteristics of the construction scenario.

6. The remote intelligent monitoring system for building construction based on 5G telephone communication and edge computing according to claim 5, characterized in that, The remote monitoring unit (400) includes a data management module (410), a visualization interaction module (420), and a command-analysis closed-loop module (430), wherein: The data management module (410) uses a hierarchical storage architecture to process edge processing results and real-time global fusion results. eigenvectors Millisecond-level read and write operations are performed using an in-memory database to ensure real-time data access efficiency; historical operating data is stored in a columnar database partitioned by time dimension, supporting fast retrieval of TB-level data, and the storage strategy is dynamically adapted to the transmission bandwidth of the 5G communication unit (300). The visualization interaction module (420) is used to construct a digital twin visualization interface for the construction scene, and transform the edge processing results into dynamic identification of equipment status, dynamic distribution trajectory of personnel and real-time curves of environmental parameters. It supports synchronous refresh of the interface of multiple terminals, and the refresh frequency is consistent with the global fusion cycle of the edge processing unit (200). The instruction-analysis closed-loop module (430) realizes the encrypted issuance of parameter adjustment instructions and historical data-driven analysis: the instructions issued to the edge processing unit (200) are encrypted using the national cryptographic SM4 algorithm and digitally signed, and the entire instruction process is recorded through blockchain evidence storage technology; at the same time, construction trend analysis is performed based on historical data, and analysis results are generated to provide data support for instruction generation, forming a closed-loop link of "instruction issuance - result feedback - trend analysis - instruction optimization".

Images