Intelligent monitoring method and system for building construction electrical equipment

Abstract

The invention belongs to the technical field of electrical equipment monitoring, and provides an intelligent monitoring method and system for building construction electrical equipment in order to solve the problems that the monitoring dimension is single and hidden dangers such as electrical fires and mechanical faults are difficult to find in advance, and the method comprises the steps that a data collection module collects electrical parameter data and images; the data processing module processes the data; the image analysis module identifies abnormal conditions; the data analysis module identifies abnormal conditions; the comprehensive evaluation module is used for comprehensively evaluating and predicting potential risks; and the intelligent decision support module optimizes a dynamic power supply strategy. According to the invention, early warning of hidden dangers such as electrical fire and mechanical faults is realized; the fault trend of the equipment is predicted, the downtime of the equipment is reduced, the fault is quickly positioned, and the power supply is quickly cut off when abnormality of the electrical equipment is detected; and the operation temperature of the equipment is maintained in a safe interval.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrical equipment monitoring, more specifically, to a building construction electrical equipment intelligent monitoring method and system. BACKGROUND

[0002] The electrical equipment of building construction refers to the electrical equipment and power system used in the construction process. These devices include cables, wires, switches, sockets, fuse boxes, distribution boxes, meters, generators, etc. The electrical equipment of building construction plays a very important role in the construction process. By providing power supply and control, it ensures the electrical safety and normal operation of the construction site. At the same time, these devices also provide necessary power conditions for construction personnel, improving construction efficiency and quality.

[0003] The prior art document with publication number CN119087019A provides an intelligent monitoring system for electrical equipment, relating to the technical field of electrical detection. By setting the fan blade and conductive layer in the air inlet pipe, the real-time detection of the internal load current of the equipment can be realized by cutting the magnetic field with the rotating fan blade. At the same time, the fan blade cooperates with the design of the first and second conductive layers, which can effectively detect the current while maintaining air flow, with double functions; a shunt channel and a separation chamber are arranged in the air inlet system, and through the ingenious design of the air inlet nozzle and the separation pipe, the air entering the equipment interior can be effectively filtered before entering. The dust and particulate matter in the air are separated by centrifugal force and collected into the dust collection tank, reducing the entry of some impurities into the equipment interior, and further improving the durability and reliability of the equipment.

[0004] The prior art solution in the above has the following defects although the related beneficial effects can be achieved by the structure of the prior art: 1. Single monitoring dimension, hidden danger early warning lag: the prior art relies on a single sensor (such as only monitoring current or temperature), which is difficult to discover complex hidden dangers such as electrical fire and mechanical failure in advance, and often only responds passively after the accident occurs. 2. Most devices only have basic overload protection, lack of hierarchical threshold setting and intelligent linkage logic, may cause large-area power failure due to local failure, or cannot timely cut off the dangerous power supply, and it is difficult to achieve precise fault isolation and accident control. 3. Lack of data-driven predictive analysis, unable to predict the trend of equipment failure in advance; fault location relies on manual troubleshooting, low efficiency, long downtime, and affects construction progress. SUMMARY

[0005] The purpose of the present application is to provide a building construction electrical equipment intelligent monitoring method and system, which solves the technical problems raised in the above background technology, realizes accurate monitoring of multi-dimensional hidden dangers of electrical equipment, captures real-time current anomalies, overheating, vibration exceeding standards and other basic fault signals through sensors, realizes early warning of electrical fire, mechanical failure and other hidden dangers, predicts equipment failure trends by analyzing historical data, reduces equipment downtime, and realizes the technical effect of fast positioning of faults by associating multi-dimensional data with knowledge graph.

[0006] The technical scheme adopted by the present application to solve the above problems is: a building construction electrical equipment intelligent monitoring method, comprising the following steps:

[0007] S1, collecting data of building construction electrical equipment;

[0008] S2, collecting electrical parameters and image data of building construction electrical equipment;

[0009] S3, processing the data collected in step S2;

[0010] S4, analyzing and identifying the images collected in step S2, and identifying abnormal conditions in time;

[0011] S5, analyzing the data collected in step S2, and identifying abnormal conditions;

[0012] S6, combining sensor data and image analysis monitoring results to comprehensively evaluate and predict potential risk problems of building construction electrical equipment;

[0013] S7, load forecasting and dynamic power supply strategy optimization;

[0014] S8, when detecting abnormal electrical equipment, quickly cutting off the power supply;

[0015] S9, ensuring that the equipment operating temperature is maintained in a safe range;

[0016] S10, when monitoring abnormal conditions or potential risks, issuing an alarm in time.

[0017] The step S3 of the present application comprises the following steps:

[0018] S31, data initialization: deploying edge computing nodes, receiving data from various sensors and cameras; performing node initialization, starting edge computing nodes, and loading core processing programs;

[0019] S32, data preprocessing: including filtering and noise reduction and data aggregation, wherein data aggregation includes:

[0020] Time synchronization: aligning the timestamps of multi-source data of the same equipment;

[0021] Feature extraction: extract feature indicators from electrical parameters and temperature data;

[0022] Data fusion: after normalization processing, the extracted features are combined into a device state vector;

[0023] S33, local decision response: preset three threshold values of early warning value, alarm value and emergency value through edge computing node; when the monitored data exceeds the threshold value, power off is executed immediately.

[0024] The step S4 of the application comprises the following steps:

[0025] S41, image data access: continuously receiving video stream data sent by the camera;

[0026] S42, image preprocessing: pre-processing the image data, including video stream frame disassembly, noise reduction processing and normalization processing;

[0027] S43, device damage and deformation detection: constructing a target detection model; inputting the pre-processed image into the target detection model, marking the area that may exist abnormally, judging whether there is deformation and damage; screening the detection results output by the target detection model; positioning the key components of the detected device, comparing the parameters of the key components with the standard installation state of the device, judging whether the device installation is stable and whether there is potential failure risk;

[0028] S44, spark detection;

[0029] S45, multi-module collaborative result output: abnormal information integration, presenting the abnormal information in a visual interface.

[0030] The step S44 of the application comprises the following steps:

[0031] S44.1, image graying: graying the pre-processed image;

[0032] S44.2, flow algorithm trajectory extraction: using Lucas-Kanade sparse optical flow algorithm to analyze the motion vector of the pixel points in the adjacent two frames of images; by tracking the motion trajectory of the pixel points, the region with fast and irregular motion characteristics is identified, and whether there is spark is preliminarily judged;

[0033] S44.3, dynamic threshold segmentation: dynamically calculating the segmentation threshold according to the overall brightness and local contrast information of the image; separating the region that may contain spark;

[0034] S44.4 Feature Extraction and Classification: For the segmented suspected spark regions, extract their geometric features, brightness features and texture features to form a feature vector, input it into a pre-trained support vector machine classifier to determine whether the region is a real spark and assess its hazard level.

[0035] S44.5 Alarm Triggering and Recording: If a spark is identified as a dangerous spark by the support vector machine classifier and its duration exceeds a specified time, an alarm mechanism will be triggered.

[0036] Step S5 of the present invention includes the following steps:

[0037] S51, Data Reception and Integration: Acquire real-time data from each sensor and calibrate the timestamps of all received data;

[0038] S52. Data preprocessing: Preprocess the collected data, including handling missing values, removing outliers, and normalizing the data.

[0039] S53, In-depth data analysis;

[0040] S54. Anomaly Identification and Confirmation: Mark suspected abnormal data records, and conduct secondary confirmation of the initially marked anomalies by combining other relevant sensor data, historical data and equipment operating conditions;

[0041] S55. Abnormal Response Handling: For confirmed abnormal situations, trigger the corresponding level of alarm according to the severity of the abnormality.

[0042] Step S53 of the present invention includes the following steps:

[0043] S53.1 Threshold Judgment Analysis: Compare and analyze real-time data with preset normal threshold ranges;

[0044] S53.2 Trend Prediction Analysis: Based on historical data, predict the future trend of data changes; if the prediction result shows that a certain parameter is about to exceed the normal range, it is considered a potential anomaly.

[0045] S53.3 Correlation Analysis: Analyze the correlation between data from different types of sensors;

[0046] S53.4 Spectrum Analysis: Convert the time-domain signal collected by the vibration sensor into a frequency-domain signal, analyze the frequency components of the equipment vibration, and identify the characteristic frequencies of mechanical faults.

[0047] Step S6 of the present invention includes the following steps:

[0048] S61. Data Acquisition: Acquire real-time data collected by various sensors; receive image analysis and monitoring results; retrieve historical operating data of electrical equipment;

[0049] S62. Data preprocessing: Preprocess the acquired data;

[0050] S63. Feature Extraction: Extract key features from sensor data, encode image analysis and monitoring results, and convert equipment anomaly type and location information into quantifiable feature vectors.

[0051] S64. Comprehensive evaluation of equipment operating status: Establish an evaluation index system; determine the weight of each evaluation index; calculate the score of electrical equipment on each evaluation index based on the preprocessed data and the set weights.

[0052] S65. Potential Risk Prediction: Construct a prediction model, using preprocessed multi-source data as input and the future fault state or performance degradation degree of electrical equipment as output labels; perform fault trend prediction to determine whether there are potential fault risks and the time points when faults may occur;

[0053] S66. Rapid fault location includes the following steps:

[0054] S66.1 Knowledge Graph Construction: Collect structural information, component relationships, failure modes and causes of electrical equipment, and construct a knowledge graph;

[0055] S66.2 Data Association Mapping: Map real-time monitoring data and anomaly information to a knowledge graph to find related nodes and paths;

[0056] S66.3 Fault Reasoning: Based on the knowledge graph structure and relationships, graph algorithms and reasoning rules are used to analyze the possible causes and impact range of faults caused by abnormal data, so as to achieve rapid and accurate fault location.

[0057] S67. Output Results: Based on the comprehensive evaluation results, potential risk predictions, and fault location information, generate a detailed equipment operation status evaluation report and output it in a visual format.

[0058] Step S7 of the present invention includes the following steps:

[0059] S71. Data Integration: Retrieve multi-source information such as electrical equipment operating parameters, environmental data, construction progress data, and historical electricity consumption data; clean the collected data and unify the data format.

[0060] S72. Load forecasting: Select a load forecasting model, train the selected model using historical electricity consumption data, evaluate the model performance through cross-validation, input the real-time collected data into the trained model, and predict the electricity load in the future period.

[0061] S73, Dynamic Power Supply Strategy Optimization:

[0062] S73.1 Develop power supply strategy rules: Develop a power supply strategy rule library based on the construction schedule, power grid load, and equipment operating status;

[0063] S73.2 Strategy Generation and Evaluation: Based on load forecasting results and power supply strategy rules, generate multiple power supply strategy schemes; simulate and evaluate each scheme.

[0064] S73.3 Optimal Strategy Selection and Execution: Select the power supply strategy with the best overall performance, send the instruction to the relevant equipment controller or power dispatching system to achieve dynamic power supply adjustment;

[0065] S74. Construct an equipment energy consumption model and automatically adjust the operating parameters of electrical equipment;

[0066] S75. Effect Evaluation: Real-time evaluation of the actual effects of load forecasting, power supply strategy and equipment parameter adjustment; timely adjustment of load forecasting model parameters, optimization of power supply strategy rules and improvement of energy consumption model based on evaluation results.

[0067] In step S8 of this invention, the power supply is quickly cut off through a multi-level linkage mechanism. The linkage logic of the multi-level linkage mechanism is as follows:

[0068] Level 1 warning: When the temperature exceeds the warning value or the current exceeds 110% of the rated value, a warning signal will be issued first, but the power supply will not be cut off;

[0069] Level 2 alarm: If the temperature exceeds the alarm value or there is a continuous overload, the power supply to the device will be automatically cut off.

[0070] Level 3 Emergency: Upon detection of sparks, smoke, or short-circuit current, the area-level power supply is cut off, and the fire suppression system is activated.

[0071] An intelligent monitoring system for electrical equipment in building construction, used to implement the above method, includes:

[0072] Data collection module: Collects data on electrical equipment used in building construction and annotates the data;

[0073] Data acquisition module: Collects electrical parameter data from electrical equipment;

[0074] Communication module: used for data transmission;

[0075] Data processing module: processes the collected data;

[0076] Image analysis module: Analyzes and identifies the acquired images to detect anomalies;

[0077] Data analysis module: Analyzes the data collected from various sensors and identifies anomalies;

[0078] Comprehensive assessment module: Combines the monitoring results of various sensor data and image analysis modules to comprehensively assess the operation of electrical equipment and predict potential risks;

[0079] Intelligent decision support module: performs load forecasting and dynamic power supply strategy optimization; automatically adjusts equipment operating parameters based on energy consumption models;

[0080] Alarm module: When an abnormal situation, potential problem or risk is detected, an alarm will be issued in a timely manner;

[0081] Data storage module: Stores and manages the collected data;

[0082] Automatic power-off protection device: quickly cuts off the power supply when an electrical equipment malfunction is detected;

[0083] Heat dissipation and cooling devices: used for equipment heat dissipation;

[0084] Control Center: Network connected to the data collection module, data acquisition module, comprehensive evaluation module, communication module, data processing module, image analysis module, data analysis module, automatic power-off protection device, heat dissipation and cooling device, data storage module, and alarm module.

[0085] Compared with the prior art, the present invention has the following advantages and effects:

[0086] 1. This invention enables precise monitoring of multi-dimensional potential hazards in electrical equipment. It uses high-precision sensors (such as fiber optic current transformers and infrared temperature sensors) to capture basic fault signals such as abnormal current, overheating, and excessive vibration in real time. Combined with AI image analysis, it enables early warning of potential hazards such as electrical fires and mechanical failures.

[0087] 2. The spark monitoring sensor can capture arc discharge within 1ms. Combined with the SVM algorithm, it can distinguish between normal operation and fault sparks, thus avoiding short circuit accidents caused by arcs.

[0088] 3. Implement multi-level linkage power failure protection: preset three-level thresholds trigger different responses: when there is over-temperature or over-pressure, an early warning will be issued first, and if the abnormality continues, the power supply to the equipment / area will be cut off in stages to prevent the fault from escalating.

[0089] 4. Data-driven predictive maintenance: By analyzing historical data, predicting equipment failure trends transforms "reactive maintenance" into "prevention," reducing equipment downtime. Knowledge graphs link multi-dimensional data to enable rapid fault location.

[0090] 5. The heat dissipation and cooling device dynamically adjusts the fan speed / liquid cooling flow rate through a PID algorithm to ensure that the equipment temperature is maintained within a safe range, avoiding insulation aging or component damage caused by overheating. Attached Figure Description

[0091] Figure 1 This is a flowchart illustrating the intelligent monitoring method for electrical equipment in building construction according to an embodiment of the present invention.

[0092] Figure 2 This is a schematic diagram of the data processing flow of the image analysis module in the intelligent monitoring system for electrical equipment in building construction, according to an embodiment of the present invention.

[0093] Figure 3 This is a schematic diagram of the data analysis module processing data in the intelligent monitoring system for electrical equipment in building construction, according to an embodiment of the present invention.

[0094] Figure 4 This is a schematic diagram illustrating the comprehensive evaluation process of the comprehensive evaluation module of the intelligent monitoring system for electrical equipment in building construction according to an embodiment of the present invention. Detailed Implementation

[0095] The present invention will be further described in detail below with reference to the accompanying drawings.

[0096] I. The intelligent monitoring system for electrical equipment in building construction according to an embodiment of the present invention includes:

[0097] (a) Data collection module, which collects data on electrical equipment used in building construction, including but not limited to:

[0098] Equipment operating parameters: Actively collect basic operating data from equipment such as power distribution cabinets, transformers, and welding machines, and obtain raw parameters by connecting to the equipment's built-in sensors or communication interfaces;

[0099] Third-party system data: By connecting with external platforms such as smart construction site management platforms and power dispatching systems through API interfaces, relevant data such as construction progress and power grid load can be obtained to enrich the dimensions of data analysis;

[0100] Manual data entry: A data entry interface is reserved to support maintenance personnel in manually entering non-real-time data such as equipment maintenance records and temporary power applications to improve equipment file information.

[0101] (II) Data Acquisition Module: Real-time acquisition of electrical parameter data of electrical equipment. The data acquisition module includes infrared temperature sensors, cameras, humidity sensors, high-precision voltage sensors, fiber optic current transformers, vibration sensors, etc. The positions and quantities of various sensors and cameras are reasonably arranged according to the number and location of electrical equipment.

[0102] Fiber optic current transformers and high-precision voltage sensors acquire parameters such as current, voltage, power, and frequency in real time; they support multiple protocol conversions including Modbus and CAN, adapting to different equipment interfaces such as distribution cabinets, transformers, and welding machines. The fiber optic current transformers utilize the Rogowski coil principle, featuring high precision (0.2S level) and wide-band response characteristics, acquiring AC / DC current signals in real time and supporting multiple protocol conversions (Modbus RTU / TCP, CANopen), adapting to the current monitoring needs of equipment such as distribution cabinets, transformers, and welding machines. The high-precision voltage sensors, based on resistive voltage division or electromagnetic induction technology, achieve a measurement accuracy of ±0.1% FS, monitoring three-phase voltage, frequency, and harmonic parameters in real time, and outputting data via RS485 interface or Ethernet.

[0103] Infrared temperature sensor: Deployed in heat-prone parts such as motor windings, cable joints, and switch contacts, it adopts non-contact temperature measurement technology, with a temperature measurement range of -20℃ to 1500℃ and an accuracy of ±2℃. It supports multi-point array installation and generates real-time thermal imaging maps of equipment temperature.

[0104] Vibration sensors: For rotating equipment such as motors, pumps, and fans, piezoelectric or accelerometer sensors are used to monitor parameters such as vibration velocity, acceleration, and displacement. FFT spectrum analysis is used to identify mechanical faults such as bearing wear and rotor imbalance. This helps monitor potential mechanical failures.

[0105] A humidity sensor detects ambient humidity to prevent a decline in insulation performance.

[0106] Smoke sensors and cameras are used for dual monitoring of electrical fire hazards. Photoelectric smoke sensors are deployed in areas with concentrated electrical equipment and cable trays, in conjunction with cameras (4K resolution, supporting AI visual analysis). Through a dual mechanism of smoke concentration threshold detection and image recognition, early warning of electrical fires can be achieved.

[0107] Spark monitoring sensors capture transient anomalies such as arc discharge in real time. Employing a combination of ultraviolet and infrared detection technology, it can capture spark signals generated by arc discharge within 1ms. Combined with AI algorithms, it distinguishes between normal operating sparks and faulty sparks, accurately locating the discharge position.

[0108] (III) Communication Module: A hybrid communication architecture combining wireless and wired transmission schemes is adopted to ensure reliable data transmission in complex construction environments.

[0109] Wireless transmission solution: High-speed data backhaul is achieved through 4G / 5G networks in ground areas; low-power wide area network technologies such as LoRa and NB-IoT are used in underground and signal-obstructed areas; short-range networking is achieved, and ZigBee self-organizing network connects distributed sensor nodes.

[0110] Wired transmission solution: Ethernet connects the data center to the large equipment controller; supports multi-protocol conversion and edge computing through industrial-grade gateways.

[0111] (iv) Data processing module: Processes the collected data; deploys edge computing nodes to realize data preprocessing and local decision-making.

[0112] Data preprocessing: real-time filtering and noise reduction to eliminate outliers caused by electromagnetic interference; achieving millisecond-level sampling of key parameters (such as overcurrent and spark signals).

[0113] Filtering and noise reduction: An adaptive Kalman filter algorithm is used to remove abnormal data caused by electromagnetic interference and noise signals in real time; for transient fault data (such as overcurrent and spark signals), the sampling frequency is increased to ensure complete waveform recording.

[0114] Data aggregation: Time synchronization and feature fusion of multi-source data (such as current, voltage, and temperature) from the same device to reduce data redundancy and reduce transmission pressure.

[0115] Local decision response: Preset thresholds trigger control commands such as emergency cut-off; edge caching ensures that data is not lost when the network is interrupted.

[0116] Threshold-triggered control: Multiple thresholds (early warning value, alarm value, emergency value) are preset. When the monitored data exceeds the threshold, the edge computing node immediately executes local control commands, such as cutting off the power to the faulty equipment or starting emergency lighting.

[0117] Edge caching mechanism: Built-in large-capacity flash memory supports 10 days of local data caching; automatically stores data when the network is interrupted, and resumes transmission after recovery to ensure data integrity.

[0118] (v) Image analysis module: Analyzes and identifies the acquired images to promptly detect abnormalities (including equipment damage, deformation, sparks, etc.).

[0119] Target detection: The YOLOv5 algorithm is used to identify abnormalities in the appearance of electrical equipment (such as deformed casing or loose wiring), with a detection accuracy of ≥95%; key components of the equipment are located through key point detection technology, and the installation status is monitored in real time.

[0120] Flame and smoke recognition: It integrates convolutional neural networks (CNN) and transfer learning techniques to quickly identify flame and smoke features with a false alarm rate of <0.1%; combined with time series analysis, it distinguishes between real fires and environmental interference.

[0121] Spark detection: Based on optical flow algorithm and dynamic threshold segmentation, the spark trajectory is extracted; the SVM classifier is used to determine the spark hazard, and an alarm is triggered for sparks with a duration of >500ms.

[0122] (vi) Data analysis module: Analyze the data collected from various sensors and identify abnormal situations in a timely manner.

[0123] (vii) Comprehensive Assessment Module: This module combines data from various sensors and monitoring results from image analysis modules to comprehensively assess the operational status of electrical equipment and predict potential risks. It uses deep learning models (LSTM, Transformer) to predict equipment failure trends and leverages knowledge graphs to link multi-dimensional data for rapid fault location. The comprehensive assessment module also dynamically optimizes power outage thresholds and cooling modes (e.g., reducing fan speed to save energy during low-load nighttime periods) by combining data such as temperature, load, and fault history.

[0124] (viii) Intelligent Decision Support Module: Performs load forecasting and dynamic power supply strategy optimization; automatically adjusts equipment operating parameters based on energy consumption models. The intelligent decision support module adjusts equipment operating parameters (such as welding machine duty cycle) in advance based on load forecasting results to avoid overload caused by concentrated heat generation.

[0125] (ix) Alarm module: including an alarm device that promptly issues an alarm when abnormal situations or potential problems and risks are detected.

[0126] (x) Data storage module: Stores and manages the collected data; uses a distributed file system to store massive amounts of monitoring data; and uses blockchain technology to ensure device identity authentication and data security.

[0127] Furthermore, data storage employs a dual-storage backup approach, performing dual backups through two different paths. Even if one path's storage backup encounters an unexpected problem, the other path can still effectively perform backups, preventing data loss.

[0128] Choose two completely independent and physically isolated storage paths. For example, one path can be based on the storage resources of a local server cluster deployed in the enterprise's own data center, equipped with high-performance disk arrays and stable network connections; the other path uses cloud storage services, such as Alibaba Cloud Object Storage (OSS) or Amazon S3. During data writing, the system simultaneously sends the data to both different storage paths. Parallel data writing can be achieved by writing specific storage middleware, which interacts with both local and cloud storage to ensure that data is accurately stored in both paths simultaneously. To ensure complete data consistency between the two storage paths, periodic data consistency checks are required. A hash check is used to calculate hash values ​​for the data in both storage paths and then compare the hash values. If the hash values ​​are different, it indicates inconsistency, requiring further investigation and processing. When inconsistency is detected, the system automatically retrieves the correct data from the normal storage path and overwrites the problematic storage path to ensure data consistency between the two paths.

[0129] The advantage of dual storage is improved data reliability and avoidance of single points of failure. In traditional single-storage-path setups, data loss is possible if a storage device fails (e.g., hard drive failure, server downtime). However, with dual storage backup, even if one storage path experiences an unexpected problem, the other path can still provide a complete data backup, significantly reducing the risk of data loss.

[0130] (xi) Automatic power-off protection device: When an electrical equipment abnormality is detected (such as overload, short circuit, overheating, fire hazard, etc.), the power supply is quickly cut off through a multi-level linkage mechanism to prevent the accident from escalating. At the same time, it supports remote / local dual control to ensure construction safety.

[0131] (xii) Heat dissipation and cooling device: In response to the heat generation problem of construction electrical equipment (such as transformers, welding machines, and distribution cabinets) under high load operation, intelligent temperature control and multi-mode heat dissipation are used to ensure that the operating temperature of the equipment is maintained within a safe range (such as the internal temperature of the distribution cabinet ≤ 40℃, and the transformer oil temperature ≤ 85℃).

[0132] (xiii) Control Center: Connected to the data collection module, data acquisition module, comprehensive evaluation module, communication module, data processing module, image analysis module, data analysis module, data storage module and alarm module via network.

[0133] II. The intelligent monitoring method for electrical equipment in building construction according to an embodiment of the present invention includes the following steps:

[0134] S1. The data collection module collects data on electrical equipment used in building construction and labels the data.

[0135] S2, the data acquisition module collects electrical parameters and image data of electrical equipment in building construction in real time; the data storage module stores and manages the collected data.

[0136] S3. The data processing module processes the data collected in step S2. Specifically, it includes the following steps:

[0137] S31. Data Initialization: Deploy edge computing nodes to perform data preprocessing and local decision-making. Edge computing nodes receive data from various sensors (such as fiber optic current transformers and infrared temperature sensors) and cameras via communication modules. Data formats include Modbus, CAN, and video streams. Node initialization is performed, starting the edge computing node and loading core processing programs such as adaptive Kalman filtering algorithms, data aggregation programs, and threshold control rules. Simultaneously, a large-capacity flash memory is initialized for data caching.

[0138] S32. Data Preprocessing: Including filtering and noise reduction, and data aggregation.

[0139] S32.1, Filtering and Noise Reduction:

[0140] S32.1.1 Signal Analysis: Perform AD conversion and protocol analysis on the received analog signals such as current, voltage, and vibration to restore them to the original physical quantity data.

[0141] S32.1.2 Kalman Filtering: Establish a state model and construct state transition equations and observation equations (such as dynamic current change models) based on historical electrical parameter data and physical characteristics. Perform prediction and updating: predict the data state at the next moment through the prediction steps of Kalman filtering; update the data by combining actual measurement values, calculate the optimal estimate, and eliminate abnormal fluctuations caused by electromagnetic interference and noise.

[0142] S32.1.3, Enhanced Transient Data Sampling: Real-time monitoring of data change rate; when transient fault characteristics such as overcurrent or spark signals are detected, the sampling frequency is dynamically increased from the normal value (e.g., 100Hz) to 10kHz. A high-speed cache is enabled to store high-frequency sampled data in a circular queue to ensure complete recording of fault waveforms.

[0143] S32.2, Data Aggregation:

[0144] S32.2.1 Time Synchronization: Based on the high-precision clock of the edge computing node (such as the GPS timing module), the timestamps of multiple sources such as current, voltage and temperature of the same device are aligned, and the error is controlled within ±1ms.

[0145] S32.2.1 Feature Extraction: Calculate and extract feature information such as effective value, peak value, and harmonic content from electrical parameters; extract key feature information such as hot spots and temperature gradients from temperature data.

[0146] S32.2.3 Data Fusion: After normalizing the extracted features, they are merged into device state vectors (such as current features, voltage features, and temperature features) to reduce the amount of data transmission.

[0147] S33, Local Decision Response, including:

[0148] S33.1 Threshold Trigger Control:

[0149] Threshold settings: Preset multiple threshold levels in the edge computing node configuration interface:

[0150] Warning value: Approaching the upper limit of the normal range, triggering a yellow warning.

[0151] Alarm value: If the value exceeds the normal range, an audible and visual alarm will be triggered and a message will be sent to the maintenance personnel.

[0152] Emergency value: When the danger threshold is reached, immediately execute control commands such as power outage protection and activation of emergency lighting.

[0153] Real-time judgment and execution: The edge computing node performs threshold comparison on the processed data every second. When the monitored data exceeds the threshold, the control command execution module is immediately invoked to cut off the power to the faulty equipment via a relay control switch, or to send a start signal to the lighting system.

[0154] S33.2, Edge caching mechanism:

[0155] Cache initialization: Dedicated cache space is allocated in the large-capacity flash memory of the edge computing node, a circular storage strategy is adopted, and a 72-hour data storage cycle is set.

[0156] Data caching and synchronization: Processed data is written to the cache in real time, with each data entry accompanied by a timestamp and device identifier. When the network is normal, cached data is synchronized to the cloud data center every 5 minutes, and synchronized data is deleted. When a network interruption is detected, it automatically switches to offline caching mode to continuously store new data; after the network is restored, cached data is resumed in chronological order to ensure that data is not lost or duplicated.

[0157] The data processing module enables efficient processing of collected data and local intelligent decision-making, ensuring the data quality and timely response of the building construction electrical equipment monitoring system.

[0158] S4. The image analysis module analyzes and identifies the acquired images, promptly detecting anomalies (including equipment damage, deformation, sparks, etc.). Specifically, it includes the following steps:

[0159] S41. Image Data Access: Connect the 4K resolution cameras deployed at the construction site to the image analysis module via Ethernet or a wireless communication module. Configure parameters such as video stream transmission protocol (e.g., RTSP), frame rate, and resolution to ensure stable image data transmission. The image analysis module continuously receives video stream data from the cameras, temporarily stores it in a local buffer, forming a continuous image data queue to provide a data foundation for subsequent analysis.

[0160] S42. Image Preprocessing: Preprocessing image data, including video stream frame splitting, noise reduction, and normalization.

[0161] Video stream splitting: Read the video stream from the buffer, split it into single-frame images in chronological order, and generate an image sequence so that each frame can be analyzed independently;

[0162] Noise reduction: For single-frame images, median filtering or Gaussian filtering algorithms are used to remove salt-and-pepper noise, Gaussian noise, etc., improving image clarity. For example, for images affected by construction site dust, median filtering can effectively smooth noise.

[0163] Normalization: The image size is uniformly scaled to the size adapted to the algorithm model, and the image pixel values ​​are normalized to map the pixel values ​​to the [0,1] interval, which speeds up the algorithm processing speed and improves the accuracy of feature extraction.

[0164] S43. Equipment damage and deformation detection:

[0165] S43.1, Building Module: Using the YOLOv5 model, start the YOLOv5 object detection model in the image analysis module, and load the weight file that has been pre-trained on the building construction electrical equipment dataset. This model has learned the appearance features of the equipment in normal and abnormal states.

[0166] S43.2 Feature Extraction and Recognition: The preprocessed image is input into the YOLOv5 object detection model. The model automatically extracts features from the image, such as device outlines and component shapes, through a multi-layer convolutional neural network. Anchor frames are used to mark areas that may be abnormal, determining whether there is deformation or damage such as casing deformation, loose wiring, or missing components.

[0167] S43.3 Detection Result Filtering: Set a confidence threshold (e.g., 0.5) to filter the detection results output by the model. Remove detection boxes with confidence levels below the threshold, retain high-confidence anomaly detection information, and record the detected anomaly type (e.g., "shell deformation"), location coordinates, and other data.

[0168] S43.4 Critical Point Status Assessment: For the detected equipment, critical point detection technology is used to locate the critical components of the equipment (such as screws, interfaces, solder joints, etc.). By calculating parameters such as the positional offset and angular change of the critical components, and comparing them with the standard installation state of the equipment, it is determined whether the equipment installation is stable and whether there are any potential fault risks.

[0169] S44. Spark Detection:

[0170] S44.1 Image Grayscale Conversion: The preprocessed image is converted to grayscale to highlight the brightness information of the image, reduce the amount of data processing, and enhance the contrast between the spark and the background.

[0171] S44.2, Flow Algorithm Trajectory Extraction: The Lucas-Kanade sparse optical flow algorithm is used to analyze the motion vectors of pixels in two adjacent frames. By tracking the motion trajectories of pixels, regions with rapid and irregular motion characteristics are identified, and the presence of sparks is preliminarily determined. The Lucas-Kanade sparse optical flow algorithm model is as follows:

[0172] I x u+I y v+I t =0; Introducing spatiotemporal window constraints, we construct a regularized optimization problem:

[0173] E(u,v)=Σ (x,y)∈Ω [(I x u+I y v+I t ) 2 +a 2 (|▽u| 2 +|▽v| 2 ]; where Ω is the local window, a is the smoothing parameter, and the optical flow field (u,v) is solved by minimizing the energy function. I x I represents the gradient of the image in the x-direction, and I represents the rate of change of image brightness in the horizontal direction; y I represents the gradient of the image in the y-direction, and I represents the rate of change of image brightness in the vertical direction; t The gradient of the image in the time direction t represents the rate of change of image brightness over time; it is used to reflect the change in pixel brightness between two adjacent frames. u is the optical flow component of the pixel in the x-direction (horizontal direction), that is, the horizontal displacement velocity of the pixel between two adjacent frames. v is the optical flow component of the pixel in the y-direction (vertical direction), that is, the vertical displacement velocity of the pixel between two adjacent frames; the optical flow information of the pixel can be obtained by solving this equation.

[0174] The spark motion feature recognition model is as follows:

[0175] M(x,y)=SQT[u 2 (x,y)+v 2 [(x,y)]; When (M(x,y)>T) M

[0176] And D(x,y)>T D When the coordinates are (x,y), it is determined to be a suspected spark region. In the formula, M(x,y) represents the motion amplitude of the pixel at coordinates (x,y) in the optical flow field, i.e., the speed at which the point moves between two frames. u(x,y) represents the optical flow component of pixel (x,y) along the horizontal x-axis, representing the horizontal movement speed. v(x,y) represents the optical flow component of pixel (x,y) along the vertical y-axis, representing the vertical movement speed. D(x,y) is the rate of change of the motion amplitude (M(x,y)) in space, reflecting the degree of difference in motion amplitude between adjacent pixels. T M This is the motion amplitude threshold; typical value: 5-10 pixels / frame (depending on the camera frame rate and spark size). In construction scenarios, such as those involving welding machines where sparks travel at high speeds, a higher threshold (e.g., 8 pixels / frame) can be set; while small sparks from poor cable connections travel at lower speeds, requiring a lower threshold (e.g., 5 pixels / frame). D This is the threshold for the rate of change of direction (which depends on the image resolution and noise level). In construction scenarios with complex backgrounds (such as worker movement or mechanical vibration), the threshold needs to be increased to eliminate interference.

[0177] S44.3 Dynamic Threshold Segmentation: Based on information such as overall image brightness and local contrast, a segmentation threshold is dynamically calculated. An adaptive thresholding algorithm (such as the Otsu algorithm) is used to segment the image into foreground (potential spark regions) and background, separating regions that may contain sparks. The Otsu algorithm model is as follows:

[0178] T(x,y)=T Otsu (x,y)+βH(x,y)+γC(x,y);

[0179] H(x,y)=-Σ L-1 i＝0 {p i (x,y)log2[p i (x,y)]};p i (x,y)=n i (x,y) / N;

[0180] C(x,y)=max i,j∈Ω |I i (x,y)-I j(x,y)|;where H(x,y) is the entropy value of the local region where the pixel at coordinates (x,y) in the image is located, used to measure the information richness and randomness of the region. p i (x, y) represents the probability of a pixel with gray value i appearing within a local region; n i (x,y) represents the number of pixels with gray value i within the local region, and N is the total number of pixels in the local region. L is the total number of gray levels in the image; T(x,y) is the adaptive segmentation threshold calculated by the improved algorithm, used to segment the image at position (x,y) into foreground (spark) and background. Otsu (x,y) is the threshold calculated by the traditional Otsu algorithm, based on the principle of maximizing inter-class variance, considering only grayscale distribution information. C(x,y) is the color contrast feature, measuring the degree of color difference within a local region. i (x,y) and I j (x,y) represents the color values ​​of different channels or positions within the local region Ω. β and γ are weighting coefficients used to adjust the influence of local entropy and color contrast on the final threshold, and need to be determined experimentally based on the actual scenario (typical values: β = 0.2, γ = 0.3).

[0181] S44.4 Feature Extraction and SVM Classification: For the segmented suspected spark regions, extract their geometric features (area, perimeter, circularity), brightness features (average brightness, brightness variance), and texture features (gray-level co-occurrence matrix features). These features are then combined into a feature vector and input into a pre-trained Support Vector Machine (SVM) classifier to determine whether the region is a real spark and assess its hazard level.

[0182] S44.5 Alarm Triggering and Recording: For sparks identified as dangerous by SVM and lasting longer than 500ms, an alarm mechanism is triggered. An alarm signal is sent to the system control center, and the time, location, and hazard level of the spark are recorded for subsequent querying and analysis.

[0183] S45. Multi-module collaborative result output: This feature integrates anomaly information by combining equipment damage and deformation detection results, spark detection results, and detection results from other image analysis modules (such as flame and smoke recognition modules) to form a unified anomaly information list. The anomaly information is presented in a visual interface, marking the location of the abnormal equipment, the anomaly type, and the hazard level on the control center's large screen or mobile app. Simultaneously, the anomaly information is reported in real-time to the system's comprehensive evaluation module, providing a basis for the comprehensive assessment of the electrical equipment's operating status.

[0184] S5, the data analysis module analyzes the data collected from various sensors to promptly identify anomalies. Specifically, this includes the following steps:

[0185] S51. Data Reception and Integration: The communication module receives real-time data from devices such as infrared temperature sensors, high-precision voltage sensors, fiber optic current transformers, vibration sensors, and humidity sensors. It also acquires non-real-time data such as manually entered equipment maintenance records and temporary power applications. For Modbus, CAN, and other protocol data output from different sensors, a data conversion tool is used to unify them into a standard format, ensuring efficient data processing by the module. The timestamps of all received data are calibrated to ensure consistency in the data's time dimension, with errors controlled within milliseconds.

[0186] S52. Data Preprocessing: Preprocess the collected data, including handling missing values, removing outliers, and normalizing the data.

[0187] Missing value handling: For cases where data is missing, methods such as linear interpolation and mean imputation are used to estimate and fill missing values ​​based on data from previous and subsequent time points to ensure data integrity.

[0188] Outlier removal: Using the 3σ principle (three standard deviations rule), obvious abnormal data points caused by sensor failure or electromagnetic interference are identified and removed to reduce their impact on the analysis results.

[0189] Data normalization: Map sensor data with different dimensions and value ranges to a unified interval through linear transformation and other methods to facilitate subsequent analysis and calculation.

[0190] S53, In-depth Data Analysis:

[0191] S53.1 Threshold Judgment Analysis: Compare real-time data with preset normal threshold ranges. For example, if the current data exceeds 1.2 times the rated current, the equipment temperature is higher than 80℃, or the ambient humidity is greater than 90%RH, it is initially judged as abnormal data.

[0192] S53.2 Trend Forecasting Analysis: Based on historical data, algorithms such as moving averages and exponential smoothing are used to predict future trends in the data. If the prediction results indicate that a certain parameter is about to exceed the normal range, it is considered a potential anomaly.

[0193] S53.3 Correlation Analysis: Analyze the correlation between data from different types of sensors. For example, check whether voltage and temperature also change accordingly when current suddenly increases, and determine whether the abnormality is caused by a single factor or multiple factors.

[0194] S53.4 Spectrum Analysis: For time-domain signals collected by vibration sensors, Fourier transform and other methods are used to convert them into frequency-domain signals, analyze the frequency components of equipment vibration, and identify characteristic frequencies of mechanical faults such as bearing wear and rotor imbalance.

[0195] S54. Anomaly Identification and Confirmation: Based on the above analysis, suspected anomaly data records are marked, generating a preliminary anomaly data list containing information such as anomaly type, occurrence time, and involved equipment. The initially marked anomalies are then reconfirmed by combining other relevant sensor data, historical data, and equipment operating conditions. For example, for temperature anomalies, data such as current and ventilation status at the same time are checked to rule out misjudgments.

[0196] S55. Anomaly Response Handling: For confirmed anomalies, trigger the corresponding alarm level (e.g., early warning, alert, emergency alarm) from the alarm module based on the severity of the anomaly, and push the anomaly information to the mobile terminal of maintenance personnel or the control center's large screen. Store the anomaly data and its related information (e.g., correlated data over a period of time) in the data storage module to form a complete anomaly event archive, facilitating subsequent fault analysis and tracing.

[0197] S6. The comprehensive evaluation module combines data from various sensors (including infrared temperature sensors, cameras, humidity sensors, high-precision voltage sensors, fiber optic current transformers, vibration sensors, etc.) with image analysis monitoring results from the image analysis module to comprehensively evaluate the operating status of electrical equipment and predict potential risks. Specifically, it includes the following steps:

[0198] S61. Data Acquisition: Acquires real-time data from infrared temperature sensors, humidity sensors, high-precision voltage sensors, fiber optic current transformers, vibration sensors, etc., from the data collection module, covering information such as equipment temperature, ambient humidity, electrical parameters, and mechanical vibration. Receives monitoring results output from the image analysis module, including equipment appearance anomalies (such as casing deformation, loose wiring), flame and smoke identification results, and spark detection information. Retrieves historical operating data, fault records, and maintenance information of electrical equipment from the data storage module as a reference for comprehensive evaluation.

[0199] S62. Data Preprocessing: Preprocessing the acquired data; including data cleaning and data alignment.

[0200] Data cleaning: The received sensor data and image analysis monitoring results are checked to remove duplicate, erroneous or incomplete data records to ensure data quality.

[0201] Data alignment: Align multi-source data based on timestamps to ensure that sensor data and image analysis results at the same moment can be accurately correlated, and to build a complete snapshot of the device status.

[0202] S63. Feature Extraction: Perform feature engineering on sensor data to extract key features, such as the effective value, peak value, and harmonic content of current, and the rate of temperature change. Encode the image analysis and monitoring results, converting information such as equipment anomaly type and location into quantifiable feature vectors.

[0203] S64. Comprehensive assessment of equipment operating status:

[0204] S64.1 Establish an evaluation index system: Develop evaluation indexes covering dimensions such as electrical performance, mechanical condition, environmental impact, and appearance integrity, such as voltage stability, vibration intensity, and the degree of influence of humidity on insulation.

[0205] S64.2 Weight Allocation: The weights of each evaluation index are determined using methods such as the Analytic Hierarchy Process (AHP) or the entropy weight method, reflecting the degree of influence of different factors on the equipment operating status.

[0206] S64.3 Comprehensive Score Calculation: Based on the preprocessed data and set weights, the scores of the electrical equipment on each evaluation indicator are calculated, and the comprehensive operating status score of the equipment is obtained by summarizing the scores, which intuitively displays the current health status of the equipment. The comprehensive evaluation model is as follows:

[0207] S 综 =KΣ M i＝1 [D i w i f(X i )];Σ M i＝1 w i =1; K=K0A; A=e -λT ;

[0208] D i ={1, v i ∈[a i ,b i ];1+ζ|v i -u i | / σ i ,v i ≠[a i ,b i ]};

[0209] f(X i )={(10X i 2 ) / (x i 2 +25), X i <5; 10-(X) i -10) 2 / 10,5≤X i ≤10; 10(10 / X)i ),X i >10};

[0210] In the formula, S 综 This is the overall equipment operating status score, with a value range of [0, 100]. A higher value indicates a better equipment status; i is the evaluation index number, i = 1, 2, ..., M; M is the total number of evaluation indicators; X i This is the raw score of the i-th evaluation indicator, calculated using different methods depending on the indicator type; w i is the basic weight of the i-th evaluation indicator, determined by the analytic hierarchy process or entropy weight method. K is the equipment characteristic correction coefficient, reflecting the comprehensive impact of equipment type, importance, and aging degree on the score; K0 is the basic coefficient of equipment type, for example: high-voltage distribution cabinet: K0 = 1.2 (critical equipment, high scoring benchmark); ordinary welding machine: K0 = 1.0 (conventional equipment, normal scoring benchmark); standby transformer: K0 = 0.8 (non-critical equipment, low scoring benchmark); A is the equipment aging coefficient; T is the equipment service life (years); λ is the aging rate constant, for example: transformer: λ = 0.05 (faster aging); distribution cabinet: λ = 0.03 (moderate aging); low-voltage switch: λ = 0.01 (slower aging); D i It is the dynamic weight adjustment factor for the i-th indicator, used to amplify the impact of abnormal indicators; f(X) i ) is a non-linear mapping function that maps the original score X to... i Mapping to a non-linear score enhances sensitivity to outliers; v i It is the real-time measurement value of the i-th indicator; [a i ,b i ] represents the normal range for the i-th indicator; u i σ is the historical average of the i-th indicator; i ζ is the historical data standard deviation of the i-th indicator; ζ is the sensitivity coefficient, which takes different values ​​for different indicator types.

[0211] S65. Potential Risk Prediction: Construct a Transformer prediction model, using preprocessed multi-source data as input and the future fault state or performance degradation degree of electrical equipment as the output label. Train the model using historical data, optimizing model parameters to enable it to learn the time-series features and complex relationships in the data. Perform fault trend prediction by inputting real-time data into the trained deep learning model to predict the performance change trend of the equipment over a future period, determining whether there are potential fault risks and the possible time points of fault occurrence.

[0212] S66. Rapid Fault Location:

[0213] S66.1 Knowledge Graph Construction: Collect knowledge such as the structural information of electrical equipment, component relationships, failure modes and causes, and construct a knowledge graph. Nodes in the knowledge graph include equipment components, failure types, influencing factors, etc., and edges represent the relationships between nodes, such as "component-failure" and "failure-cause".

[0214] S66.2 Data Association Mapping: Map real-time monitoring data and anomaly information to a knowledge graph to find related nodes and paths.

[0215] S66.3 Fault Reasoning: Based on the structure and relationships of knowledge graphs, graph algorithms and reasoning rules are used to analyze the possible causes and impact range of faults caused by abnormal data, so as to achieve rapid and accurate fault location.

[0216] S67. Output Results: Based on the comprehensive assessment results, potential risk predictions, and fault location information, a detailed equipment operation status assessment report is generated, including the current status, risk warnings, and fault recommendations. The assessment report and warning information are pushed to the control center and mobile devices of maintenance personnel. Simultaneously, the equipment's operating status and risk distribution are displayed intuitively on a visual interface, facilitating timely understanding of the equipment's condition and enabling appropriate measures to be taken by relevant personnel.

[0217] S7, the intelligent decision support module performs load forecasting and dynamic power supply strategy optimization; it automatically adjusts equipment operating parameters based on energy consumption models. Specifically, it includes the following steps:

[0218] S71. Data Integration: Retrieve multi-source information from the data collection module, including electrical equipment operating parameters (current, voltage, power, etc.), environmental data (temperature, humidity), construction progress data, and historical electricity consumption data. Clean the collected data to remove duplicate, erroneous, or invalid data; standardize the data format to ensure data accuracy and consistency.

[0219] S72. Load Forecasting: Based on the characteristics of building construction and data features, select an appropriate load forecasting model, such as a time series analysis model, a machine learning regression model (random forest regression, gradient boosting tree), or a deep learning model (LSTM, Transformer). Train the selected model using historical electricity consumption data, adjust model parameters to improve prediction accuracy, and evaluate model performance through methods such as cross-validation to ensure model reliability and generalization ability. Input the real-time collected data into the trained model to predict electricity load over a future period, providing a basis for power supply strategy optimization.

[0220] S73, Dynamic Power Supply Strategy Optimization:

[0221] S73.1 Develop power supply strategy rules: Develop a power supply strategy rule base based on the construction schedule, power grid load, and equipment operating status. For example, prioritize power supply to critical equipment during peak construction periods, and appropriately reduce the power of non-critical equipment when the power grid load is tight.

[0222] S73.2 Strategy Generation and Evaluation: Based on load forecasting results and power supply strategy rules, generate multiple power supply strategy schemes; use simulation technology to simulate and evaluate each scheme, and compare the advantages and disadvantages of the schemes from multiple dimensions such as power supply stability, cost, and energy consumption.

[0223] S73.3 Optimal Strategy Selection and Execution: Select the power supply strategy with the best overall performance, and send the instructions to the relevant equipment controller or power dispatching system through the communication module to realize dynamic power supply adjustment.

[0224] Score j =w1S 稳 +w2(1 / C 成 )+w3E total ;

[0225] S 稳 =1-(1 / nc)Σ nc i＝1 |(V i,act -V i,related ) / V i,related |;w1+w2+w3=1;

[0226] C 成 =Σ T1 t＝1 (P t E t )+Σ nc t＝1 (k i L i ); E total =Σ T1 t＝1 Σ nc t＝1 (P i，t E t In the formula, Score j S is the comprehensive score of the j-th power supply strategy, calculated by weighting indicators such as power supply stability, cost, and energy consumption. A higher comprehensive score indicates a better strategy, used to select the optimal power supply strategy from multiple options. w1, w2, and w3 are weighting coefficients, corresponding to power supply stability, cost, and energy consumption indicators, which can be determined using multi-objective decision-making methods such as the Analytic Hierarchy Process (AHP) or the entropy weight method. 稳This is a power supply stability index, with values ​​ranging from 0 to 1. The closer the value is to 1, the higher the power supply stability. Power supply stability is assessed by calculating the deviation rate between the actual supply voltage and the rated voltage of each device. `nc` represents the total number of devices and is used to calculate the average voltage deviation rate of all devices, reflecting the overall power supply stability. V i,act V represents the actual power supply voltage of the i-th device, a data obtained through real-time monitoring, reflecting the actual power supply status of the device during operation. i,related is the rated voltage of the i-th device, which is the standard voltage parameter for normal operation of the device, used to compare with the actual power supply voltage to calculate the deviation. C 成 This represents the total cost of implementing the power supply strategy, consisting of electricity costs and equipment depreciation costs, and is used to evaluate the economic cost of the strategy. T1 is the total time within the statistical period (e.g., one day, one month), used to calculate the electricity costs and equipment operating time during that period. P i，t Let be the power of the i-th device at time t. By summing the power of all devices at each time, we obtain the total power at different times. Then, by summing over the entire statistical period, we obtain the total energy consumption. E total It represents the total energy consumption during the implementation of the power supply strategy, used to evaluate the energy consumption of the strategy, and is an important indicator for measuring the energy-saving effect of the strategy. t This represents the electricity price at time t. Electricity prices may vary at different times, and the cost of electricity is determined by the power consumption. i L is the loss coefficient of the i-th device, reflecting the loss characteristics of the device itself, and is one of the parameters for calculating the loss cost of the device. i P is the runtime of the i-th device, the cumulative runtime of the device within the statistical period, which, together with the loss coefficient, determines the device loss cost. t It is the power consumption at time t, reflecting the power load of the entire power supply system at different times, and is one of the key parameters for calculating electricity costs.

[0227] S74. Automatic Adjustment of Equipment Operating Parameters Based on Energy Consumption Model: Analyze the operating characteristics and energy consumption patterns of different electrical equipment, establish an equipment energy consumption model, and clarify the relationship between equipment operating parameters (such as motor speed and welding machine current intensity) and energy consumption. The equipment operating parameters are acquired in real time through a data acquisition module and compared with the standard parameters in the energy consumption model to determine whether the equipment is operating at high efficiency.

[0228] Automatic parameter adjustment: When the equipment operating parameters are found to deviate from the optimal value, resulting in increased energy consumption, the intelligent decision support module automatically calculates and generates adjustment instructions. The control module then fine-tunes the equipment parameters to bring the equipment back to an energy-saving operating state.

[0229] S75. Effectiveness Evaluation: Continuously monitor equipment operating status, power supply conditions, and energy consumption data; evaluate the actual effectiveness of load forecasting, power supply strategies, and equipment parameter adjustments in real time. Based on the evaluation results, adjust load forecasting model parameters, optimize power supply strategy rules, and improve energy consumption models in a timely manner to continuously enhance the accuracy and effectiveness of the intelligent decision support module, thereby achieving efficient and energy-saving operation of building electrical equipment.

[0230] S8. When an electrical equipment malfunction is detected (such as overload, short circuit, overheating, fire hazard, etc.), the automatic power-off protection device quickly cuts off the power supply through a multi-level linkage mechanism to prevent the accident from escalating.

[0231] Furthermore, the automatic power-off protection device includes: intelligent circuit breaker, arc fault detection module (AFDD), remote control interface and emergency power-off controller.

[0232] The intelligent circuit breaker adopts a molded case type (such as the ABB Tmax XT series), integrating short-circuit protection, overload protection, and ground fault protection functions, and supports communication with the monitoring system via the Modbus protocol. It is deployed at the incoming end of distribution cabinets, the outgoing end of transformers, and the front end of high-power equipment such as welding machines to achieve hierarchical protection.

[0233] Key parameters: Breaking capacity ≥50kA (suitable for high current surges in construction scenarios); Action time ≤20ms (rapid disconnection during short circuit faults); Supports undervoltage / overvoltage protection (triggered when voltage fluctuations ±10%).

[0234] Emergency power outage controller: An independent control unit based on a microprocessor (MCU) that supports receiving trip commands from edge computing nodes (such as RS485 signals or hardwired signals); equipped with a manual emergency stop button to meet on-site emergency operation needs; and built-in backup power supply (lithium battery, with a battery life of ≥2 hours) to ensure that power outage commands can still be executed when the network is interrupted.

[0235] The linkage logic of the multi-level linkage mechanism is as follows:

[0236] Level 1 warning: When the temperature exceeds the warning value (e.g., cable joint temperature ≥ 70℃) or the current exceeds 110% of the rated value, a warning signal will be issued first, but the power supply will not be cut off.

[0237] Level 2 alarm: If the temperature exceeds the alarm value (≥85℃) or there is a continuous overload (exceeding the rated value by 120% and lasting for 5 minutes), the power supply to the equipment (such as a single welding machine) will be automatically cut off.

[0238] Level 3 Emergency: Upon detection of sparks, smoke, or short-circuit current, the area-level power supply (such as the entire distribution cabinet) is cut off within 0.1 seconds, and the fire protection system is activated.

[0239] Arc Fault Detection Module (AFDD): The dedicated Arc Fault Detection Module (AFDD) is used to identify series / parallel arc faults by detecting current waveform distortion. With a sensitivity of ≥99%, it can distinguish between normal switching arcs and faulty arcs, reducing false trips.

[0240] Remote control interface: Supports receiving remote power-off commands from the control center via 4G / 5G network, compatible with web and mobile terminal operations; has operation log recording function, which can trace the cause of power outage and operation time.

[0241] The status of the automatic power-off protection device (such as circuit breaker opening and closing signals, fault type) is uploaded to the control center in real time via the communication module and dynamically displayed on the monitoring interface.

[0242] Furthermore, the heat dissipation and cooling device includes: an intelligent temperature control system and an active heat dissipation module;

[0243] Intelligent temperature control system: Real-time temperature monitoring with an accuracy of ±1℃ is achieved through infrared temperature sensors (deployed at the heat source) and thermocouples (embedded inside the equipment); The control logic is as follows: Set multiple temperature control thresholds (e.g., warning value: 60℃, start heat dissipation; alarm value: 75℃, strengthen heat dissipation; emergency value: 90℃, force power off); Automatically adjust the heat dissipation intensity based on PID algorithm to avoid frequent start-stop damage to the equipment.

[0244] Active cooling modules include forced air cooling and liquid cooling systems.

[0245] Forced air cooling equipment: Install low-noise axial flow fans (such as Delta AFB1212SH, air volume ≥120CFM) in the power distribution cabinet and welding machine cabinet. The fan speed is dynamically adjusted according to the temperature (2000-4000rpm). The air inlet is equipped with a dust filter (filtration accuracy ≥5μm) to prevent construction dust from entering the equipment.

[0246] Liquid cooling equipment: For high-power transformers (capacity ≥ 500kVA), a closed-loop liquid cooling system (such as 3M Novec coolant) is adopted. The oil temperature is reduced by plate heat exchangers and external cooling fans, and the heat dissipation efficiency is improved by 40%. Flow sensors and pressure sensors are deployed to monitor the liquid cooling circuit status in real time. In case of leakage, an alarm is automatically triggered and the power is cut off.

[0247] Furthermore, the equipment cabinet is made of aluminum alloy (thermal conductivity ≥200W / m·K) to increase the area of ​​the heat dissipation fins; high thermal conductivity silicone grease (thermal conductivity ≥5W / m·K) is applied to the cable joints to reduce contact resistance heating. Phase change heat dissipation materials (such as paraffin-based composite materials) are filled in key parts such as transformer windings and motor stators to absorb heat through the phase change process and slow down the rate of temperature rise.

[0248] Based on meteorological data of the construction site (such as ambient temperature and humidity), adjust the heat dissipation strategy in advance: automatically increase the fan speed to the highest level in high temperature weather (≥35℃); activate the dehumidification and heating function in humid environments (heating cable power ≥50W / m) to prevent condensation from affecting insulation.

[0249] The operating parameters of the heat dissipation and cooling device (such as fan speed and coolant temperature) are connected to the data analysis module to predict the equipment life and the efficiency of the heat dissipation system.

[0250] S9. The heat dissipation and cooling device ensures that the equipment operating temperature is maintained within a safe range.

[0251] S10. When an abnormal situation or potential problem or risk is detected, the alarm module will issue an alarm in a timely manner.

[0252] The application of the technical solution of the present invention is illustrated by specific examples, specifically embodiment 1:

[0253] Scenario: Cluster monitoring of electrical equipment during construction of a super high-rise office building:

[0254] Project Background: A construction project for a 300-meter super high-rise office building has deployed more than 300 electrical equipment, including tower cranes (2 units), construction elevators (4 units), welding machines (20 units), distribution cabinets (15 units), and transformers (3 units). The peak daily electricity consumption is 50,000 kWh.

[0255] 1. Equipment deployment and data acquisition:

[0256] Fiber optic current transformer: installed in the tower crane distribution box, it monitors the three-phase current (rated value 120A), voltage (380V) and harmonic content in real time with an accuracy of 0.2S.

[0257] Infrared temperature sensor: deployed in a 5×5 array at the welding machine cable joint, with a temperature measurement range of -20℃ to 1500℃ and an accuracy of ±2℃.

[0258] Vibration sensor: embedded in the bearing of the construction elevator motor, sampling frequency 10kHz, monitoring vibration acceleration (threshold ≤5g).

[0259] 2. Key experimental data:

[0260] Overload warning response: In August 2024, during tower crane hoisting, the current surged to 168A (140% of the rated 120A). The edge computing node triggered an early warning within 0.5 seconds, and the intelligent decision-making module simultaneously reduced the welding machine's duty cycle to prevent power grid overload. Compared to traditional manual inspection, the fault detection time was reduced by 90%.

[0261] Temperature anomaly handling: When the temperature of the welding machine cable joint reaches 92℃ (safe threshold 85℃), the forced air cooling system (fan speed 4000rpm) is activated, and the temperature drops to 75℃ within 10 minutes. The heat dissipation response time is ≤10 seconds, and the temperature control accuracy is ±3℃.

[0262] Energy saving effect: The dynamic power supply strategy reduces peak load fluctuations by 15%, saves 12,000 yuan in electricity costs per month, and reduces equipment failure rate by 40% year-on-year. Specific Implementation Example 2:

[0264] Scenario: Electrical fire warning during subway tunnel construction:

[0265] Project Background: Construction of a 5-kilometer subway tunnel, with ambient humidity often reaching 95% RH, deployment of power distribution cabinets (8 units), ventilation fans (12 units) and lighting system, using a LoRa+5G hybrid communication architecture.

[0266] 1. Equipment deployment and data acquisition:

[0267] Ultraviolet-infrared composite spark sensor: One unit is deployed every 50 meters along the cable tray. It can capture arc signals within 1ms and distinguish between normal operation sparks and fault sparks.

[0268] 4K camera: Combined with YOLOv5 algorithm, it detects deformation of the device casing (accuracy ≥95%), and simultaneously links with photoelectric smoke sensor (threshold 0.15dB / m).

[0269] Humidity sensor: One unit is deployed every 100 meters in the tunnel to monitor the ambient humidity (accuracy ±3% RH), and a heat tracing cable (50W / m) is used for moisture protection.

[0270] 2. Key experimental data: Early warning of fire: In March 2025, the oxidation of cable joints produced a spark that lasted for 600ms. The system identified the spark within 0.8 seconds using an optical flow algorithm and an SVM classifier, cutting off the power supply to the area 30 seconds earlier than traditional smoke alarms. The spark recognition accuracy rate was 98.7%.

[0271] Moisture-proof insulation protection: The heating tape will automatically start when the humidity is >90% RH, and the humidity inside the distribution cabinet will be maintained below 60% RH, with the insulation resistance stable at 10MΩ (safety threshold ≥0.5MΩ).

[0272] Communication reliability: At a depth of 2 kilometers underground, the LoRa transmission packet loss rate is <0.5%, and the latency is ≤500ms, meeting the requirements for real-time monitoring. Please refer to Appendix 1 for specific data.

[0273] Appendix 1:

[0274]

[0275]

[0276] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent monitoring of electrical equipment in building construction, characterized in that, Includes the following steps: S1. Collect data on electrical equipment used in building construction; S2. Collect electrical parameters and image data of electrical equipment used in building construction; S3. Process the data collected in step S2; S4. Analyze and identify the images acquired in step S2 to promptly identify any abnormalities. S5. Analyze the data collected in step S2 and identify any abnormal situations; S6. Combine the data from various sensors and the results of image analysis and monitoring to comprehensively assess the operation of electrical equipment in building construction and predict potential risks. S7. Perform load forecasting and dynamic power supply strategy optimization; S8. When an electrical equipment malfunction is detected, quickly disconnect the power supply; S9. Ensure that the equipment operating temperature is maintained within a safe range; S10. When an abnormal situation or potential risk is detected, an alarm shall be issued in a timely manner.

2. The intelligent monitoring method for electrical equipment in building construction according to claim 1, characterized in that: Step S3 includes the following steps: S31. Data Initialization: Deploy edge computing nodes to receive data from various sensors and cameras; perform node initialization, start the edge computing nodes, and load the core processing program; S32. Data preprocessing: This includes filtering and noise reduction, and data aggregation, where data aggregation includes: Time synchronization: Timestamp alignment of multi-source data from the same device; Feature extraction: Feature indicators are extracted from electrical parameters and temperature data; Data fusion: After normalizing the extracted features, they are merged into a device state vector; S33. Local decision response: Preset three-level thresholds (early warning, alarm, and emergency) through edge computing nodes; immediately cut off power when the monitored data exceeds the threshold.

3. The intelligent monitoring method for electrical equipment in building construction according to claim 1, characterized in that: Step S4 includes the following steps: S41, Image Data Access: Continuously receives video stream data sent by the camera; S42. Image preprocessing: Preprocessing image data, including video stream frame splitting, noise reduction and normalization. S43. Equipment damage and deformation detection: Construct a target detection model; input the preprocessed image into the target detection model, mark areas that may be abnormal, and determine whether there is deformation or damage; filter the detection results output by the target detection model; for the detected equipment, locate the key components of the equipment, and compare the calculated parameters of the key components with the standard installation state of the equipment to determine whether the equipment installation is stable and whether there are potential fault risks. S44, Spark Detection; S45. Multi-module collaborative output: Integrate abnormal information and present it in a visual interface.

4. The intelligent monitoring method for electrical equipment in building construction according to claim 3, characterized in that: Step S44 includes the following steps: S44.1 Image Grayscale Conversion: Perform grayscale conversion on the preprocessed image; S44.2 Flow Algorithm Trajectory Extraction: The Lucas-Kanade sparse optical flow algorithm is used to analyze the motion vectors of pixels in two adjacent frames; by tracking the motion trajectory of pixels, regions with fast and irregular motion characteristics are identified, and the presence of sparks is preliminarily determined. S44.3 Dynamic Threshold Segmentation: Dynamically calculate the segmentation threshold based on the overall brightness and local contrast information of the image; separate out the regions that may contain sparks; S44.4 Feature Extraction and Classification: For the segmented suspected spark regions, extract their geometric features, brightness features and texture features to form a feature vector, input it into a pre-trained support vector machine classifier to determine whether the region is a real spark and assess its hazard level. S44.5 Alarm Triggering and Recording: If a spark is identified as a dangerous spark by the support vector machine classifier and its duration exceeds a specified time, an alarm mechanism will be triggered.

5. The intelligent monitoring method for electrical equipment in building construction according to claim 1, characterized in that: Step S5 includes the following steps: S51, Data Reception and Integration: Acquire real-time data from each sensor and calibrate the timestamps of all received data; S52. Data preprocessing: Preprocess the collected data, including handling missing values, removing outliers, and normalizing the data. S53, In-depth data analysis; S54. Anomaly Identification and Confirmation: Mark suspected abnormal data records, and conduct secondary confirmation of the initially marked anomalies by combining other relevant sensor data, historical data and equipment operating conditions; S55. Abnormal Response Handling: For confirmed abnormal situations, trigger the corresponding level of alarm according to the severity of the abnormality.

6. The intelligent monitoring method for electrical equipment in building construction according to claim 5, characterized in that: Step S53 includes the following steps: S53.1 Threshold Judgment Analysis: Compare and analyze real-time data with preset normal threshold ranges; S53.2 Trend Prediction Analysis: Based on historical data, predict the future trend of data changes; if the prediction result shows that a certain parameter is about to exceed the normal range, it is considered a potential anomaly. S53.3 Correlation Analysis: Analyze the correlation between data from different types of sensors; S53.4 Spectrum Analysis: Convert the time-domain signal collected by the vibration sensor into a frequency-domain signal, analyze the frequency components of the equipment vibration, and identify the characteristic frequencies of mechanical faults.

7. The intelligent monitoring method for electrical equipment in building construction according to claim 1, characterized in that: Step S6 includes the following steps: S61. Data Acquisition: Acquire real-time data collected by various sensors; receive image analysis and monitoring results; retrieve historical operating data of electrical equipment; S62. Data preprocessing: Preprocess the acquired data; S63. Feature Extraction: Extract key features from sensor data, encode image analysis and monitoring results, and convert equipment anomaly type and location information into quantifiable feature vectors. S64. Comprehensive evaluation of equipment operating status: Establish an evaluation index system; determine the weight of each evaluation index; calculate the score of electrical equipment on each evaluation index based on the preprocessed data and the set weights. S65. Potential Risk Prediction: Construct a prediction model, using preprocessed multi-source data as input and the future fault state or performance degradation degree of electrical equipment as output labels; perform fault trend prediction to determine whether there are potential fault risks and the time points when faults may occur; S66. Rapid fault location includes the following steps: S66.1 Knowledge Graph Construction: Collect structural information, component relationships, failure modes and causes of electrical equipment, and construct a knowledge graph; S66.2 Data Association Mapping: Map real-time monitoring data and anomaly information to a knowledge graph to find related nodes and paths; S66.3 Fault Reasoning: Based on the knowledge graph structure and relationships, graph algorithms and reasoning rules are used to analyze the possible causes and impact range of faults caused by abnormal data, so as to achieve rapid and accurate fault location. S67. Output Results: Based on the comprehensive evaluation results, potential risk predictions, and fault location information, generate a detailed equipment operation status evaluation report and output it in a visual format.

8. The intelligent monitoring method for electrical equipment in building construction according to claim 1, characterized in that: Step S7 includes the following steps: S71. Data Integration: Retrieve multi-source information such as electrical equipment operating parameters, environmental data, construction progress data, and historical electricity consumption data; clean the collected data and unify the data format. S72. Load forecasting: Select a load forecasting model, train the selected model using historical electricity consumption data, evaluate the model performance through cross-validation, input the real-time collected data into the trained model, and predict the electricity load in the future period. S73, Dynamic Power Supply Strategy Optimization: S73.1 Develop power supply strategy rules: Develop a power supply strategy rule library based on the construction schedule, power grid load, and equipment operating status; S73.2 Strategy Generation and Evaluation: Based on load forecasting results and power supply strategy rules, generate multiple power supply strategy schemes; simulate and evaluate each scheme. S73.3 Optimal Strategy Selection and Execution: Select the power supply strategy with the best overall performance, send the instruction to the relevant equipment controller or power dispatching system to achieve dynamic power supply adjustment; S74. Construct an equipment energy consumption model and automatically adjust the operating parameters of electrical equipment; S75. Effect Evaluation: Real-time evaluation of the actual effects of load forecasting, power supply strategy and equipment parameter adjustment; timely adjustment of load forecasting model parameters, optimization of power supply strategy rules and improvement of energy consumption model based on evaluation results.

9. The intelligent monitoring method for electrical equipment in building construction according to claim 1, characterized in that: In step S8, the power is quickly cut off through a multi-level linkage mechanism. The linkage logic of the multi-level linkage mechanism is as follows: Level 1 warning: When the temperature exceeds the warning value or the current exceeds 110% of the rated value, a warning signal will be issued first, but the power supply will not be cut off; Level 2 alarm: If the temperature exceeds the alarm value or there is a continuous overload, the power supply to the device will be automatically cut off. Level 3 Emergency: Upon detection of sparks, smoke, or short-circuit current, the area-level power supply is cut off, and the fire suppression system is activated.

10. An intelligent monitoring system for electrical equipment in building construction, used to implement the method according to any one of claims 1-9, characterized in that: include: Data collection module: Collects data on electrical equipment used in building construction and annotates the data; Data acquisition module: Collects electrical parameter data from electrical equipment; Communication module: used for data transmission; Data processing module: processes the collected data; Image analysis module: Analyzes and identifies the acquired images to detect anomalies; Data analysis module: Analyzes the data collected from various sensors and identifies anomalies; Comprehensive assessment module: Combines the monitoring results of various sensor data and image analysis modules to comprehensively assess the operation of electrical equipment and predict potential risks; Intelligent decision support module: performs load forecasting and dynamic power supply strategy optimization; automatically adjusts equipment operating parameters based on energy consumption models; Alarm module: When an abnormal situation, potential problem or risk is detected, an alarm will be issued in a timely manner; Data storage module: Stores and manages the collected data; Automatic power-off protection device: quickly cuts off the power supply when an electrical equipment malfunction is detected; Heat dissipation and cooling devices: used for equipment heat dissipation; Control Center: Network connected to the data collection module, data acquisition module, comprehensive evaluation module, communication module, data processing module, image analysis module, data analysis module, automatic power-off protection device, heat dissipation and cooling device, data storage module, and alarm module.

Images