Traffic engineering construction site electricity safety intelligent management and control system

Abstract

The invention relates to the technical field of traffic engineering construction safety management and control, and discloses a traffic engineering construction site power utilization safety intelligent management and control system, which comprises a sensing layer, a transmission layer, an edge computing layer, a cloud platform layer and an application layer, and each layer realizes data interaction through an MQTT protocol. According to the traffic engineering construction site power utilization safety intelligent management and control system, an environment-electrical parameter linkage early warning module and a data preprocessing module of an edge calculation layer improve power utilization early warning accuracy, and a sensing layer collects environment temperature and humidity, dust concentration and electrical parameters and then transmits the environment temperature and humidity, the dust concentration and the electrical parameters to the edge calculation layer through double links of a transmission layer; according to the method, abnormal value filtering and normalization are completed through the data preprocessing module, then the linkage early warning module calculates a comprehensive early warning coefficient according to a coefficient model, an electric parameter safety threshold value can be dynamically adjusted, for example, an electric leakage early warning threshold value is lowered to 25 mA when the humidity exceeds 85% RH, the false alarm problem of a single parameter fixed threshold value is avoided, and early warning is made to be more suitable for complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of safety management and control technology for traffic engineering construction, specifically to an intelligent management and control system for electrical safety at traffic engineering construction sites. Background Technology

[0002] Traffic engineering construction sites are often characterized by a wide variety of electrical equipment, drastic fluctuations in construction load, scattered and complex temporary power supply lines, and harsh working environments such as open-air and special conditions. Therefore, the safe management of temporary power supply has become a core and critical link in the safety management system of traffic engineering construction and is a fundamental prerequisite for ensuring the smooth progress of the project.

[0003] Currently, although the management of temporary power consumption at transportation engineering construction sites has gradually introduced technical means such as basic electrical parameter monitoring equipment, simple early warning devices, and conventional communication transmission links, most projects will install monitoring instruments such as ammeters and voltmeters in distribution boxes at all levels. Some projects are equipped with single threshold leakage and overload early warning systems. At the same time, they rely on 4G public networks or wired networks to realize the remote transmission of power consumption data, thereby achieving basic supervision of the power consumption status on site.

[0004] However, existing power management technologies for traffic engineering construction sites still have many significant shortcomings and are difficult to adapt to the complex construction conditions of traffic engineering projects: First, the accuracy of electricity warnings is insufficient. Most existing monitoring systems only set fixed threshold alarms for a single electrical parameter, such as only monitoring whether the current exceeds the rated value, but ignoring the associated impact that rainy and high-humidity environments will exacerbate the risk of leakage, resulting in frequent false alarms and safety officers becoming complacent about the warning signals. Second, there is a lack of traceability of hazard data. The entire process from hazard discovery to rectification is mostly recorded on paper or verbally, without encrypted evidence storage and closed-loop links. For example, after the hazard of loose wiring in the distribution box is rectified, it is impossible to provide comparative data before and after rectification and the signature of the responsible person for evidence storage, making it difficult to pass the compliance verification of the safety supervision department. Third, the emergency response capability is weak. When cloud communication is interrupted, the site cannot independently execute basic safety strategies. For example, in one project, due to cloud disconnection, the power distribution box failed to automatically shut off after being overloaded, causing local line overheating faults. At the same time, the access control is chaotic, and there is a risk of misoperation by the team leader in violation of regulations by operating the main power distribution box. Therefore, we proposed an intelligent management and control system for electrical safety at transportation engineering construction sites. Summary of the Invention

[0005] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides an intelligent management and control system for electrical safety at traffic engineering construction sites. This system has advantages such as accurate backup power early warning, traceable hidden dangers, and efficient emergency response. It solves the problems of insufficient accuracy in electrical early warning, lack of traceability of hidden danger data, and weak emergency response capabilities in existing electrical management and control technologies for traffic engineering construction sites.

[0006] (II) Technical Solution To achieve the aforementioned goals of accurate backup power early warning, traceable hidden dangers, and efficient emergency response, this invention provides the following technical solution: an intelligent management and control system for power safety at traffic engineering construction sites, comprising a perception layer, a transmission layer, an edge computing layer, a cloud platform layer, and an application layer, with each layer interacting with data via the MQTT protocol; The sensing layer includes an electrical parameter acquisition module, an environmental parameter acquisition module, and an equipment status acquisition module. The electrical parameter acquisition module is deployed in various levels of distribution boxes and the power supply end of large construction machinery to collect current, voltage, leakage current, and cable joint temperature data. The environmental parameter acquisition module is used to collect ambient temperature, humidity, and dust concentration data. The equipment status module is used to collect the start-stop status of construction machinery and the open / closed status of distribution boxes. The transmission layer adopts a dual-mode networking structure of 4G / 5G main communication link and LoRa backup communication link, including an edge gateway with built-in breakpoint resume module and LoRa repeater. When the main link is interrupted, it automatically switches to the backup link and caches data locally. After the link is restored, the cached data is uploaded according to the timestamp. The edge computing layer is deployed in the computing chip of the edge gateway and has a built-in data preprocessing module and an environment-electrical parameter linkage early warning module, which are used to preprocess the data collected by the perception layer and realize local hierarchical early warning. The cloud platform layer is deployed on a cloud server and integrates an AI fault diagnosis model and a hidden danger data traceability module. It is used to receive data from the edge computing layer and realize full-link data storage for fault diagnosis and hidden danger handling. The application layer includes a monitoring screen, a mobile terminal APP, and a compliance report generation module, which are used to push early warning information, visualize electricity consumption data, and output compliance reports.

[0007] Preferably, the data preprocessing module of the edge computing layer performs outlier filtering and data normalization. Outlier filtering calculates the mean and standard deviation of electrical parameters over 10 consecutive acquisition cycles, identifying and removing values ​​that deviate from the mean by three times the standard deviation. The data normalization formula is as follows: In the formula This is the original collected data. This is the lower limit for parameter monitoring. This represents the upper limit for parameter monitoring.

[0008] Preferably, the environmental-electrical parameter linkage early warning module of the edge computing layer has a built-in early warning coefficient calculation model, and the model formula is as follows: In the formula To provide a comprehensive early warning coefficient, This is the humidity weighting factor (value 0.4). This is the temperature weighting coefficient (value 0.3). This is the dust weighting coefficient (value 0.3). This represents the ambient humidity value. This refers to the temperature value of the cable connector. This refers to the dust concentration value. The module dynamically adjusts the electrical parameter safety threshold according to environmental conditions. When the ambient humidity is >85%RH, the leakage current warning threshold is lowered to 25mA. When the cable joint temperature is >80℃ and the dust concentration is >10mg / m³, a level 2 warning is triggered.

[0009] Preferably, the AI ​​fault diagnosis model of the cloud platform layer is a BP neural network model. In the model construction phase, the input layer is set to have 5 nodes corresponding to normalized data of current, voltage, leakage current, cable temperature, and ambient humidity, respectively. The hidden layer has 2 layers with 12 and 8 nodes, respectively. The output layer has 6 nodes corresponding to 6 types of electrical faults: short circuit, leakage, overload, over-temperature, voltage instability, and current fluctuation. In the model training phase, 10,000 sets of historical fault data are used, with cross-entropy as the loss function. The model is iteratively trained using the Adam optimizer at a learning rate of 0.001 until the convergence accuracy is 0.001. The model completes self-iterative optimization with an iteration cycle of 30 days. After the model is deployed, it receives preprocessed data from the edge computing layer and outputs fault diagnosis results. The fault identification accuracy of the updated model is not less than 95%.

[0010] Preferably, the hazard data traceability module of the cloud platform layer constructs a full-chain data link for hazard handling, which includes full-node data such as hazard discovery time, hazard type, responsible person for notification, handling time, handling measures, and review results; the module uses a hash encryption algorithm to encrypt the data, and the encryption formula is as follows: In the formula For encrypted digest, Original data for hazard mitigation. The encrypted data is stored in a distributed storage architecture in association with the original data, with a storage period of 5 years. Each data entry is associated with a unique electronic signature to achieve full-chain traceability.

[0011] Preferably, the electrical parameter acquisition module of the sensing layer has a current monitoring accuracy of no less than 0.5A, a voltage monitoring accuracy of no less than 0.1V, and a leakage current monitoring resolution of no more than 0.01mA, with an acquisition frequency of 1 second / time; the environmental parameter acquisition module has a temperature and humidity monitoring range covering -20℃ to 60℃, 10%RH to 95%RH, and a dust concentration monitoring resolution of no more than 0.01mg / m³, with an acquisition frequency of 5 seconds / time; the equipment status acquisition module is triggered by real-time acquisition of status changes, and the acquired data is encrypted using the national cryptographic SM4 encryption algorithm before transmission, with a transmission delay of no more than 100ms.

[0012] Preferably, the local cache capacity of the edge gateway of the transmission layer is not less than 16GB, and the communication coverage radius of the LoRa repeater is not less than 800 meters.

[0013] Preferably, the edge computing layer further includes a strategy pre-execution module, which pre-stores basic power control strategies such as overload power outage and timed power outage at night. When communication with the cloud platform layer is interrupted, the overload power outage strategy is automatically executed. The overload power outage trigger condition is that the circuit current exceeds 1.2 times the rated current for 30 seconds. The permission management module of the cloud platform layer sets three levels of permissions: Level 1 permission allows the project manager to view the power consumption data of the entire project and execute global power outage operations; Level 2 permission allows the safety officer to view the power consumption data of their assigned area and execute area power outage operations; and Level 3 permission allows the team leader to view only the team's power consumption data. Permission verification adopts a dual verification method of account password plus dynamic verification code, and the dynamic verification code is valid for 5 minutes.

[0014] Preferably, the mobile terminal APP of the application layer adopts a hierarchical early warning approach. The first-level early warning corresponds to short circuit and leakage faults, and is delivered through audible and visual alarms, SMS, and APP push notifications. The second-level early warning corresponds to overload and over-temperature faults, and is delivered through APP push notifications and telephone reminders. The third-level early warning corresponds to parameter fluctuation faults, and is delivered through APP push notifications. The early warning response time does not exceed 3 minutes. The compliance report generation module automatically generates daily / weekly / monthly electricity safety reports, and the report format complies with the filing requirements of the "Technical Specification for Temporary Electricity Safety at Construction Sites".

[0015] A method for intelligent management and control of electrical safety at traffic engineering construction sites, used to implement an intelligent management and control system for electrical safety at traffic engineering construction sites, includes the following steps: Step 1: The electrical parameter acquisition module, environmental parameter acquisition module, and equipment status acquisition module of the sensing layer acquire corresponding data respectively. After the acquisition is completed, the data is encrypted and transmitted to the transmission layer. Step 2: The transport layer transmits data through the 4G / 5G main link. When the main link is interrupted, it switches to the LoRa backup link. At the same time, the edge gateway caches the data. After the link is restored, the cached data is uploaded according to the timestamp. Step 3: The edge computing layer performs outlier filtering and normalization preprocessing on the received data, and then calculates the comprehensive early warning coefficient through the environment-electrical parameter linkage early warning module. Based on the coefficient, the corresponding level of local early warning is triggered. When cloud communication is interrupted, the preset power control strategy is executed. The environment-electrical parameter linkage early warning module automatically adjusts the leakage current early warning threshold according to the ambient humidity value. When the ambient humidity is >85%RH, the leakage current early warning threshold is lowered from 30mA to 25mA. When the cable joint temperature is >80℃ and the dust concentration is >10mg / m³, a level 2 early warning is triggered and the early warning data is uploaded to the cloud platform layer simultaneously. Step 4: The cloud platform layer receives preprocessed data from the edge computing layer, outputs the fault type through the AI ​​fault diagnosis model, and simultaneously the hidden danger data traceability module encrypts and stores the hidden danger handling data and generates an electronic signature; the self-iterative optimization process of the AI ​​fault diagnosis model is to collect 100 sets of new fault data every 30 days, supplement them to the training sample set and retrain the model, and the fault identification accuracy of the updated model is no less than 95%; Step 5: The mobile terminal APP at the application layer receives the graded early warning information, the monitoring screen displays the visualized content of electricity consumption data, and the compliance report generation module outputs the electricity safety report that meets the filing requirements.

[0016] (III) Beneficial Effects Compared with existing technologies, this invention provides an intelligent management and control system for electrical safety at traffic engineering construction sites, which has the following beneficial effects: 1. The intelligent control system for electrical safety at the construction site of this transportation project improves the accuracy of electrical warnings through the environmental-electrical parameter linkage early warning module and data preprocessing module at the edge computing layer. After the sensing layer collects environmental temperature, humidity, dust concentration and electrical parameters, they are transmitted to the edge computing layer via dual links at the transmission layer. The data preprocessing module first completes the outlier filtering and normalization, and then the linkage early warning module calculates the comprehensive early warning coefficient according to the coefficient model. It can also dynamically adjust the electrical parameter safety threshold. For example, when the humidity exceeds 85%RH, the leakage warning threshold is lowered to 25mA, avoiding the false alarm problem of fixed threshold for a single parameter, making the early warning more adaptable to complex working conditions.

[0017] 2. The intelligent management and control system for electrical safety at the construction site of this transportation project solves the problem of missing hazard traceability through the hazard data traceability module at the cloud platform layer. After the AI ​​fault diagnosis model diagnoses the fault, the traceability module will construct a link containing full node data of the hazard. The data is encrypted using the SHA256 encryption algorithm combined with timestamps, associated with a unique SM2 electronic signature, and stored in a distributed architecture for 5 years, forming a closed-loop link of encrypted evidence storage. It can provide complete rectification data evidence storage and successfully pass the compliance verification of the safety supervision department.

[0018] 3. The intelligent power safety management and control system at the construction site of this transportation project enhances emergency response capabilities through the strategy pre-execution module at the edge computing layer and the permission management module at the cloud platform layer. When cloud communication is interrupted, the strategy pre-execution module automatically executes overload power-off and timed power-off strategies at night. The permission management module sets up three levels of permissions and uses double verification with account password plus dynamic verification code. Different levels correspond to different operation permissions to avoid team leaders' violations and prevent power safety failures caused by cloud connection interruptions. Attached Figure Description

[0019] Figure 1 This is a flowchart of the intelligent management and control method for electrical safety at traffic engineering construction sites according to the present invention; Figure 2 This is a diagram illustrating the architecture of the intelligent control system for electrical safety at construction sites of transportation engineering projects, as described in this invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments and 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1-2 The intelligent management and control system for electrical safety at transportation engineering construction sites includes a perception layer, a transmission layer, an edge computing layer, a cloud platform layer, and an application layer. Each layer interacts with data via the MQTT protocol. The sensing layer includes an electrical parameter acquisition module, an environmental parameter acquisition module, and an equipment status acquisition module. The electrical parameter acquisition module is deployed in various levels of distribution boxes and the power supply end of large construction machinery to collect current, voltage, leakage current, and cable joint temperature data. The environmental parameter acquisition module is used to collect ambient temperature, humidity, and dust concentration data. The equipment status module is used to collect the start-stop status of construction machinery and the open / closed status of distribution boxes. The transmission layer adopts a dual-mode networking structure of 4G / 5G main communication link and LoRa backup communication link, including an edge gateway with built-in breakpoint resume module and LoRa repeater. When the main link is interrupted, it automatically switches to the backup link and caches data locally. After the link is restored, the cached data is uploaded according to the timestamp. The edge computing layer is deployed in the computing chip of the edge gateway and has a built-in data preprocessing module and an environment-electrical parameter linkage early warning module, which are used to preprocess the data collected by the perception layer and realize local hierarchical early warning. The cloud platform layer is deployed on a cloud server and integrates an AI fault diagnosis model and a hidden danger data traceability module. It is used to receive data from the edge computing layer and realize full-link data storage for fault diagnosis and hidden danger handling. The application layer includes a monitoring screen, a mobile terminal APP, and a compliance report generation module, which are used to push early warning information, visualize electricity consumption data, and output compliance reports.

[0022] Example 1: This embodiment provides a detailed explanation of the overall hierarchical architecture of the system, the hardware installation and deployment of each module in the perception layer, and the precise data collection operation. Combined with the deployment logic of temporary power distribution equipment and intelligent monitoring terminals at the construction site, the technical solution is implemented to ensure the accuracy of the system's all-dimensional data collection and the security of its transmission.

[0023] I. System Overall Hierarchical Architecture Setup According to the system design requirements, a five-level overall architecture is built, consisting of a perception layer, a transmission layer, an edge computing layer, a cloud platform layer, and an application layer. Data interaction between each layer is achieved using the MQTT protocol. The specific hardware and software collaborative deployment process is as follows: Each layer of equipment is embedded with an MQTT communication module, configured with a unified communication topic and parameters. The communication topic is "Traffic_Construction_Electric_Safety / + / data", the communication port is 1883, and the heartbeat interval is 30 seconds. The raw data collected by the perception layer is encrypted with SM4 and then encapsulated into MQTT message packets. The edge gateway of the transport layer acts as an MQTT client to receive message packets and forward them to the edge computing layer. The integrity and real-time performance of data interaction between each layer are guaranteed by the QoS2 level of the MQTT protocol, ensuring that critical early warning data is delivered 100%.

[0024] II. Hardware Installation and Deployment Process for the Perception Layer 1. Installation and parameter configuration of electrical parameter acquisition module Installation locations: Terminals 15 to 20 centimeters from the box body at the inlet or outlet of the primary distribution box; circuit junctions of the secondary or tertiary distribution boxes; inside the distribution box at the power supply end of large construction machinery such as tower cranes or concrete pump trucks; connected in parallel with the circuit breaker. Wiring requirements: Use shielded cables with a diameter of not less than 1.5 square millimeters. The cable laying path should be far away from areas with strong electromagnetic interference. Cables in open areas should be fitted with anti-crushing metal conduits. Parameter calibration: The system was calibrated according to the designed range of acquisition parameters, including an accuracy range of 0.3 to 0.5A for current monitoring, 0.08 to 0.12V for voltage monitoring, and 0.008 to 0.012mA for leakage current monitoring. The acquisition frequency was set to 1 time per second. The temperature range of the cable connector temperature sensor was calibrated to -25℃ to 180℃. During installation, the sensor should be placed close to the metal part of the cable connector and fixed with insulating high-temperature resistant tape.

[0025] 2. Installation and parameter configuration of the environmental parameter acquisition module Installation locations: within 5 to 8 meters of the distribution box, at the four corners of the large construction machinery operation area, and at the ventilation openings of the material storage area; for open-air locations, the modules should be fitted with an IP65-rated rainproof and dustproof shell. In areas with high dust concentration, such as cement mixing areas, the module deployment density should be increased to 1 unit per 500 square meters. Installation height: The temperature and humidity sensor should be installed at a height of 2.0 to 3.0 meters, and the dust concentration sensor should be installed at a height of 1.5 to 2.0 meters, matching the height of the worker's breathing belt; Parameter calibration: The temperature and humidity monitoring range covers -20℃ to 60℃ and 10%RH to 95%RH. The dust concentration monitoring resolution is calibrated to 0.008 to 0.012 mg / m³, and the sampling frequency is set to once every 5 seconds.

[0026] 3. Installation and Encryption Configuration of Device Status Acquisition Module Installation locations: at the moving and stationary contacts of the switch contacts in the distribution box, and at the relay contacts of the start / stop control terminal of construction machinery; Encryption process: The module is pre-integrated with the national standard SM4 encryption algorithm. After installation, the encryption key is written on-site. The key length is 128 bits and the update cycle is 30 to 45 days. After the module collects the opening and closing of the distribution box and the start and stop status data of the machinery, it completes encryption and triggers transmission within 50 to 80 milliseconds. The transmission delay is controlled within the range of 80 to 100 milliseconds.

[0027] III. Data Acquisition and Operation Process of the Perception Layer After each module has been installed and calibrated, data acquisition should be performed according to the following procedure: Step 1: The electrical parameter module collects current, voltage, leakage current and cable connector temperature once per second. The collected data is converted from analog to digital, and after ADC precision of 16 bits, it is temporarily stored in the module's local cache with a capacity of not less than 128KB. Step 2: The environmental parameter module collects temperature, humidity, and dust concentration every 5 seconds. The data is then filtered to eliminate environmental interference. Step 3: The device status module adopts a status change triggered acquisition mode. When the status of the switch contact or relay changes, acquisition is immediately started and encryption is triggered. The encrypted data is transmitted synchronously to the transmission layer with data from other modules.

[0028] In this embodiment, the high-precision range calibration of the electrical parameter module, with a current range of 0.3 to 0.5A and a voltage range of 0.08 to 0.12V, can capture minute parameter fluctuations and identify slight leakage currents of 20 to 25mA, solving the problem that traditional equipment cannot identify low-level leakage currents; the wide monitoring range and high-density deployment of the environmental module are adapted to the complex environment of the construction site and avoid distortion of environmental parameter monitoring.

[0029] The device status module uses the national standard SM4 encryption, and the 128-bit key can prevent data tampering or theft. The encrypted data transmission strength is better than traditional simple encryption, which meets the requirements for secure transmission of electrical data at construction sites.

[0030] Example 2: This embodiment implements the dual-mode networking structure of the transmission layer and the data preprocessing module of the edge computing layer in the system design. Combined with the redundancy design requirements of the transmission link at the construction site, it ensures the reliability of data transmission and the accuracy of preprocessing.

[0031] I. Hardware Setup and Deployment for Dual-Link Transport Layer 1. Establishment of 4G / 5G main communication links Edge gateway installation: Select a dry and well-ventilated area in the central control room or power distribution box cluster area of ​​the construction site to install the edge gateway. The communication distance between the gateway and the sensing layer module should be controlled between 50 and 100 meters. Wall-mounted installation should be used, 1.8 to 2.2 meters above the ground. Communication module configuration: The gateway has a built-in industrial-grade 4G / 5G communication module, a dedicated network traffic card, an uplink bandwidth of not less than 10Mbps and a downlink bandwidth of not less than 20Mbps, and the gateway communication priority is set to the 4G / 5G main link. Signal enhancement: For areas with weak signal coverage at the construction site, where the signal strength is less than -95dBm, install 4G / 5G signal boosters. The boosters should be installed at a height of not less than 6 meters and have a coverage radius of not less than 200 meters.

[0032] 2. Setting up a backup communication link using LoRa LoRa repeater deployment: LoRa repeaters are deployed at intervals of 500 to 800 meters, and multiple repeaters are cascaded to achieve coverage of the construction site without dead angles; the repeaters are installed on the tower crane boom at a height of not less than 20 meters or on a dedicated pole at a height of not less than 10 meters, with a protection level of IP67. Link switching logic configuration: Preset the link switching trigger threshold in the edge gateway. When the main link signal strength is no greater than -95 to -105dBm and lasts for 4 to 6 seconds, it will automatically switch to the LoRa backup link with a switching delay of no more than 200ms.

[0033] 3. Hardware and Functional Deployment of the Resume Download Module Local cache configuration: The gateway has a built-in local storage unit with a cache capacity of 16 to 32 GB; cached data is stored in order of timestamp and data priority, with electrical parameter data having a higher priority than environmental parameters, which in turn have a higher priority than device status. Resume interrupted transmission process: When the link is interrupted, the gateway automatically writes encrypted data to the local cache; after the link is restored, the data is uploaded in the order of timestamp from earliest to latest. After the upload is completed, the cloud is confirmed to receive the data through CRC32 verification. After the verification is successful, the local data is deleted within 10 to 15 seconds to avoid redundancy.

[0034] II. Complete Process of Edge Computing Layer Hardware Deployment and Data Preprocessing 1. Edge computing hardware deployment The edge computing module is deployed on the quad-core ARM architecture computing chip of the edge gateway, with a main frequency of 1.4 to 1.6 GHz and 4 to 8 GB of memory. The communication interface between the chip and the gateway is PCIe 3.0. During deployment, a heat dissipation module is installed on the computing chip, with a cooling fan speed of 2000 to 3000 r / min to ensure long-term operational stability.

[0035] 2. Complete Data Preprocessing Workflow Design (1) Full-process implementation of outlier filtering Calculation window setting: Preset calculation window of 8 to 12 consecutive acquisition cycles; Data reading and calculation: The module reads 10 sets of continuous electrical parameter data in chronological order and calculates the mean and standard deviation of the data within the window using a sliding window algorithm; Outlier threshold: Values ​​that deviate from the mean by 2.5 to 3.5 times the standard deviation are defined as outliers; Filtering example: The mean of the current data of a certain distribution box over 10 periods is in the range of 45 to 55A and the standard deviation is in the range of 1.8 to 2.2A. If the data for a certain period is 56 to 58A, which deviates from the mean by 3.2 to 3.5 times the standard deviation, it is judged as an outlier and removed. After removal, the data is completed using the mean within the window.

[0036] (2) Full-process implementation of data normalization Normalization formula execution: Complete data mapping according to the following formula. Preset parameter threshold range: Enter the upper and lower limit ranges for monitoring each electrical parameter in the module; Current: A range of 0 to 5A. It is 95 to 105A; Voltage: 0 to 10V 380 to 420V; Leakage current: 0 to 2mA The current is 45 to 55 mA. Cable connector temperature: The temperature ranges from -25°C to -15°C. The temperature ranges from 140 to 160 degrees Celsius.

[0037] Calculation and output: After reading the raw data, the module completes the normalization calculation within 5 to 10 milliseconds, maps the data to the 0 to 1 range, and outputs it to the subsequent modules of the edge computing layer.

[0038] In this embodiment, the dual-mode networking of 4G / 5G main link plus LoRa backup link can solve the problem of main link signal interruption. Combined with 16 to 32GB of local cache, the data transmission coverage reaches 100%. The low power consumption of LoRa link, with an operating current of no more than 50mA, reduces the overall energy consumption of the transmission layer and is suitable for long-term operation requirements at construction sites.

[0039] Filtering by subtracting 2.5 to 3.5 times the standard deviation from the mean of 8 to 12 periods can effectively eliminate instantaneous fault data of sensors and avoid false alarms; normalization unifies parameters of different magnitudes into the range of 0 to 1, eliminating dimensional differences and laying a data foundation for subsequent model analysis.

[0040] Example 3: This embodiment implements the environmental electrical parameter linkage early warning module and strategy pre-execution module designed for the system, and combines them with the leakage protection requirements of the high humidity and high dust environment at the construction site to improve the accuracy of power consumption early warning and emergency response capabilities.

[0041] I. Construction, Training, and Deployment of an Environmental Electrical Parameter-Linked Early Warning Model (1) Architecture construction of early warning model Embed a comprehensive early warning coefficient calculation model in edge computing chips: In the formula, To provide a comprehensive early warning coefficient, This is the normalized ambient humidity value. This is the normalized cable connector temperature value. This represents the normalized dust concentration value. Warning threshold range setting: No warning: Y is less than 0.25 to 0.35; Level 3 warning: 0.35≤Y < 0.6; Level II warning: 0.6 ≤ Y < 0.8; Level 1 warning: Y ≥ 0.8.

[0042] (2) Model training and validation Training dataset construction: Collect 1,000 to 1,500 sets of environmental electrical parameter correlation data from the construction site over the past two years, and divide them into training set, validation set, and test set in a ratio of 7:2:1; Training parameter configuration: using mean squared error as the loss function, employing gradient descent optimizer, setting the learning rate to 0.001 to 0.003, the number of iterations to 500 to 800, and training until the convergence accuracy is no greater than 0.001; Verification results: The prediction error of the early warning coefficient of the model validation set is no greater than 5%, which meets the early warning accuracy requirements of the construction site.

[0043] (3) Edge deployment and operation of the model Deployment process: The trained model is packaged into a lightweight TensorFlow Lite model and written to the storage unit of the edge computing chip via serial port, storage address 0x00080000; Operation process: The module reads environmental parameters every 5 seconds, substitutes them into the model to calculate the Y value, and triggers the corresponding warning if the warning threshold is reached. At the same time, the electrical parameter safety threshold is dynamically adjusted according to the following logic: when the ambient humidity is greater than 80% to 90%RH, the leakage current warning threshold is lowered from the normal range of 25 to 35mA to 20 to 28mA; when the cable joint temperature is greater than 75 to 85℃ and the dust concentration is greater than 8 to 12mg / m³, a level 2 warning is immediately triggered.

[0044] II. Hardware Linkage and Triggering Process of the Strategy Pre-execution Module (1) Hardware deployment and triggering of overload power-off strategy Hardware linkage: The edge gateway is linked with the control switch of the power distribution box through the RS485 interface, and the response delay of the control switch is no more than 50ms; Triggering condition range: The loop current lasts for 25 to 35 seconds and exceeds the rated current by 1.15 to 1.25 times; Execution process: When cloud communication is interrupted, the module automatically starts the strategy, monitors the circuit current in real time, and sends a power-off command to the control switch within 50 to 100 ms after the trigger condition is met, cutting off the power supply to the overload circuit.

[0045] (2) Configuration and execution of nighttime timed power outage strategy Time period configuration: Based on the construction site's work schedule, the power outage periods are set as the intervals from 21:30 to 6:30 and from 22:30 to 6:00. Circuit differentiation: The module pre-stores a list of emergency circuits, including emergency lighting and fire-fighting equipment. Power outage commands are only issued to the distribution boxes of non-emergency power circuits to ensure power supply to critical equipment.

[0046] In this embodiment, the weight coefficient range of the linkage early warning model is designed to be 0.35 to 0.45, 0.25 to 0.35, and 0.25 to 0.35 to achieve multi-parameter comprehensive early warning. The leakage threshold is lowered to 20 to 28mA under high humidity environment, which can identify low-level leakage risks in advance and significantly reduce the false alarm rate compared with traditional systems.

[0047] The strategy pre-execution module can autonomously execute overload power cuts when the connection to the cloud is lost. It can identify overloads and cut off power within 25 to 35 seconds, avoiding line overheating faults. Timed power cuts at night can reduce the risk of power consumption during unattended periods.

[0048] Example 4: This embodiment implements the AI ​​fault diagnosis model, hidden danger data traceability module, and access control module designed for the system, combining the needs of fault diagnosis and hidden danger closed-loop management at the construction site to achieve accurate fault diagnosis and full-chain traceability of hidden dangers.

[0049] I. Construction, Training, Deployment and Operation of BP Neural Network AI Fault Diagnosis Model (1) Complete construction of model architecture Input layer: 5 neurons, corresponding to normalized data of current, voltage, leakage current, cable temperature, and ambient humidity, with a data dimension of 5 x 1; Hidden layers: There are 2 layers. The first layer has 10 to 14 neurons and the activation function is ReLU. The second layer has 6 to 10 neurons and the activation function is Sigmoid. Output layer: 6 neurons, corresponding to 6 types of electrical faults: short circuit, leakage, overload, overheating, voltage instability, and current fluctuation. The activation function is Softmax.

[0050] (2) Model training and iterative optimization Dataset construction: Collect 8,000 to 12,000 sets of historical fault data from traffic engineering construction sites over the past 5 years, label the fault types, and divide them into training set, validation set, and test set in a ratio of 7:2:1; Training parameter configuration: Loss function: Cross-entropy loss function; Optimizer: Adam optimizer, learning rate range 0.001 to 0.005; Training iterations: Training until convergence accuracy is 0.0008 to 0.0012, with training batch size ranging from 32 to 64; Validation and optimization: The accuracy of fault identification on the validation set should be no less than 92%. For fault types with an identification rate of less than 85%, such as current fluctuations, supplement with 200 to 300 sets of samples for retraining.

[0051] Self-iterative optimization process: The iteration cycle is 25 to 35 days; each cycle collects no less than 80 to 120 sets of new fault data from the construction site and adds them to the training set; after the model is retrained, the model version iteration is completed through the cloud interface, and the model recognition accuracy stabilizes in the range of 93% to 97% after the update.

[0052] (3) Cloud deployment and operation of the model Server selection: 2 cloud servers, each with an 8-core, 16-thread CPU, 32GB of memory, and a Tesla T4 GPU, using a primary / standby mode; Deployment environment: Set up a Python 3.8 and TensorFlow 2.8 environment, and develop MQTT data receiving interface and fault result push interface; Operation process: After the preprocessed data of the edge computing layer is uploaded to the cloud via the MQTT protocol, the model outputs the fault diagnosis results within 100 to 200 ms, and the results are simultaneously pushed to the application layer monitoring screen and mobile terminal.

[0053] II. Hardware Storage and Encryption Deployment of the Hidden Danger Data Traceability Module (1) Hardware construction of the entire data link for potential risks Data node configuration: Six data nodes are built in a cloud-based distributed storage architecture to record the time of hazard discovery, hazard type, responsible person for notification, handling time, handling measures, and review results. The nodes are stored using SSD hard drives with a single disk capacity of not less than 1TB and a data read / write speed of not less than 500MB / s. Link Trigger: After the AI ​​model diagnoses a potential fault, the module automatically generates a fault data record and simultaneously triggers a push notification to the responsible person via SMS interface.

[0054] (2) Complete implementation of data encryption and storage Encryption algorithm deployment: Data encryption is performed according to the following formula: Hash = SHA256(D∥T) Data splicing: The original data D of hazard handling is spliced ​​with the millisecond-level timestamp T. The length of the spliced ​​data is no more than 1024 bytes. Encryption implementation: Deploy the SHA256 encryption algorithm on a cloud server, with encryption time not exceeding 10ms; Storage and Signature Configuration: The storage period is 4 to 6 years; the electronic signature uses the national cryptographic SM2 algorithm to generate a unique electronic signature, with a key length of 256 bits, and the signature is stored in association with the encrypted data; distributed storage is adopted, with 3 data copies to ensure that the data is not lost.

[0055] III. System Setup and Configuration of the Three-Level Access Control Module The access level is three, including: Level 1 Access: The project manager can view the power consumption data for the entire project and can execute a global power outage operation. Power outage commands require dual authentication. Level 2 access: Safety officers can view the power consumption data of 1 to 3 construction zones under their jurisdiction and execute power outage operations in the area; Level 3 access control: Team leaders can only view the power consumption data of 1 to 2 work teams in their team, and do not have the authority to cut off power. Verification method: The system uses a dual verification method consisting of an account password and a dynamic verification code. The dynamic verification code is valid for 3 to 8 minutes and is generated by the cloud server using a time synchronization algorithm.

[0056] A permission management ledger is set up on the cloud platform, permissions are assigned according to job responsibilities, permission changes require online approval from the project safety management department, the approval process takes no more than 24 hours, and the ledger data is backed up every 7 days.

[0057] In this embodiment, the two-layer hidden layer structure of the BP neural network with 10 to 14 and 6 to 10 neurons and the cross-entropy loss function can fit the nonlinear relationship between power parameters and faults, with an identification accuracy of 93% to 97%, which is better than the traditional rule-based diagnostic method; the self-iterative mechanism can adapt to changes in operating conditions, solving the problem that traditional models cannot identify new faults.

[0058] SHA256 encryption and SM2 signature can prevent data tampering, distributed storage ensures data traceability and solves the problem of no evidence for rectification of hidden dangers; the full-node data link realizes closed-loop management of hidden dangers.

[0059] Example 5: This embodiment implements the application-layer mobile terminal hierarchical early warning and compliance report generation modules of the system design, and combines the early warning response and safety management standardization requirements of construction site to achieve accurate early warning delivery and automatic report generation.

[0060] I. Hardware linkage and configuration for tiered early warning systems Level 1 warning corresponds to short circuits and leakage faults; it adopts an alarm system of audible and visual alarms + SMS push + APP push. The audible and visual alarms are linked to the audible and visual alarm devices on the construction site, with an alarm intensity of 80 to 100 dB and a light flashing frequency of 2 to 3 Hz. The alarm devices are installed in prominent locations in each construction zone; SMS push is sent to the emergency phone numbers of the project manager and safety officer through a cloud SMS gateway, with a delivery delay of no more than 60 seconds; APP push is triggered by the mobile terminal APP to be pinned to the top, and the push pop-up window is forcibly displayed; Level 2 warning corresponds to overload and over-temperature faults; it adopts an alarm method of APP push + telephone reminder. The APP push is triggered by the mobile terminal APP to push high priority; the telephone reminder is automatically initiated by cloud-based voice reminder, and the reminder voice is generated by TTS engine. The Level 3 warning corresponds to a parameter fluctuation fault, which only triggers regular push notifications from the mobile terminal APP.

[0061] The warning response time threshold is set to 2 to 4 minutes in the APP. The duration of the alert is monitored by the cloud timing module. The actual response time can be controlled within the range of 1.5 to 2.5 minutes, which meets the emergency response requirements.

[0062] II. System Configuration and Operation of the Compliance Report Generation Module The report content consists of preset daily, weekly, and monthly report templates, including statistics on electricity consumption parameters, average daily current, voltage, leakage current, fault hazard statistics, number of hazards, type, and handling rate, statistics on environmental parameters, average daily temperature and humidity, and dust concentration. The report format is designed according to the electrical safety specifications for construction sites, and includes filing elements such as report number, project name, statistical period, compiler, and reviewer. It supports export in PDF and Excel formats.

[0063] The statistical period for the compliance report generation module is: Daily Report: Generated between 00:00 and 00:10 daily; Weekly report: Generated every Sunday from 00:00 to 00:20; Monthly report: Generated between 00:00 and 00:30 at the end of each month.

[0064] The compliance report generation module automatically collects data for the corresponding time period from the cloud platform, with data collection taking no more than 5 minutes. After the report is generated, it is automatically uploaded to the project safety management ledger, and manual download is also supported. The report format can be directly used for compliance verification by safety supervision departments.

[0065] In this embodiment, the tiered outreach method matches the means according to the severity of the fault. Multiple outreach for level 1 faults can ensure that the responsible person responds in a timely manner, while simple push notifications for level 3 faults avoid information redundancy. A response threshold of 2 to 4 minutes can enable rapid fault handling.

[0066] Automatic report generation reduces the compilation time from the traditional 2 hours per report to 3 to 8 minutes per report. The report content covers all dimensions of data, and the format meets the filing requirements, thus solving the problem of substandard reports.

[0067] Example 6: This embodiment implements the overall management and control method of the system throughout the entire process, integrating the technical solutions of the above embodiments to achieve end-to-end management and control from data collection to report output.

[0068] Step 1: Data Acquisition and Encrypted Transmission at the Sensing Layer: The electrical parameter module collects current from 0 to 105A, voltage from 0 to 420V, leakage current from 0 to 55mA, and cable temperature from -25℃ to 160℃ every second once; the environmental parameter module collects temperature, humidity, and dust concentration every 5 seconds once; the device status module collects data in real time when the status changes, encrypts it with the national cryptographic standard SM4, and transmits it synchronously to the transmission layer with a 128-bit key, with a transmission delay of no more than 100ms.

[0069] Step 2, Dual-link data transmission and breakpoint resumption at the transport layer: Prioritize transmission via the 4G / 5G primary link. When the primary link signal is no greater than -95dBm and lasts for 5 seconds, automatically switch to the LoRa backup link. The edge gateway has a local cache of 16 to 32GB, stores data by timestamp, and completes the upload of cached data within 5 minutes after the link is restored, achieving 100% data integrity.

[0070] Step 3: Edge computing layer data preprocessing and linkage early warning: Outliers are removed by subtracting three times the standard deviation from the mean of 10 collection cycles, and then the data is mapped to the 0 to 1 range using a normalization formula; by calculating the comprehensive early warning coefficient Y, when the ambient humidity is greater than 85%RH, the leakage threshold is reduced from 30mA to 25mA; when the cable temperature is greater than 80℃ and the dust concentration is greater than 10mg / m³, a level 2 early warning is triggered; when the cloud connection is lost, an overload power cut-off strategy is automatically implemented, which includes a current exceeding 1.2 times the rated value for 30 seconds and a timed power cut-off strategy from 22:00 to 6:00 at night.

[0071] Step 4: Fault diagnosis and hidden danger tracing at the cloud platform layer: The BP neural network model outputs the fault type, and 100 new sets of data are added every month to complete the model iteration, with an identification accuracy of no less than 95%; the hidden danger data is encrypted with SHA256 and stored for 5 years with SM2 electronic signature to achieve full-chain traceability.

[0072] Step 5: Application Layer Warning Push and Report Output: Level 1 warnings trigger audio-visual alerts, SMS, and APP push notifications; Level 2 warnings trigger APP and telephone reminders; Level 3 warnings trigger APP push notifications, with a response time of no more than 3 minutes; daily, weekly, and monthly compliance reports are automatically generated, supporting export in multiple formats and uploading for filing.

[0073] In this embodiment, the full-link control method realizes the synergy of technical solutions at all levels. The precise collection of data at the perception layer provides a foundation for edge preprocessing, the edge linkage early warning provides screening for cloud diagnosis, the cloud diagnosis provides a basis for application layer early warning, and the application layer output realizes the implementation of control. After the method was adopted in the actual test, the occurrence rate of power failure at the construction site was greatly reduced, the closed-loop rate of hidden danger handling reached 100%, and the power safety control level of the traffic engineering construction site was comprehensively improved.

[0074] In summary, this intelligent control system for electrical safety at transportation engineering construction sites improves the accuracy of electrical warnings through its environmental-electrical parameter linkage early warning module and data preprocessing module at the edge computing layer. After the sensing layer collects environmental temperature, humidity, dust concentration, and electrical parameters, the data is transmitted to the edge computing layer via dual links at the transmission layer. The data preprocessing module first performs outlier filtering and normalization, and then the linkage early warning module calculates the comprehensive early warning coefficient according to the coefficient model. It can also dynamically adjust the electrical parameter safety threshold, such as lowering the leakage current warning threshold to 25mA when the humidity exceeds 85%RH, avoiding the false alarm problem of fixed thresholds for single parameters, and making the early warning more adaptable to complex working conditions.

[0075] Furthermore, the intelligent management and control system for electrical safety at the construction site of this transportation project solves the problem of missing hazard traceability through the hazard data traceability module at the cloud platform layer. After the AI ​​fault diagnosis model diagnoses the fault, the traceability module will construct a link containing full node data of the hazard. The data is encrypted using the SHA256 encryption algorithm combined with timestamps, associated with a unique SM2 electronic signature, and stored in a distributed architecture for 5 years, forming a closed-loop link for encrypted evidence storage. This can provide complete rectification data evidence storage and successfully pass the compliance verification of the safety supervision department.

[0076] Furthermore, this intelligent power safety management and control system for transportation engineering construction sites enhances emergency response capabilities through a strategy pre-execution module at the edge computing layer and a permission management module at the cloud platform layer. When cloud communication is interrupted, the strategy pre-execution module automatically executes overload power-off and timed power-off strategies at night. The permission management module sets up three levels of permissions and uses a dual verification method of account password plus dynamic verification code. Different levels correspond to different operation permissions, avoiding team leaders' violations and eliminating power safety faults caused by cloud connection interruptions. This solves the problems of insufficient accuracy of power early warning, lack of hazard data traceability, and weak emergency response capabilities in existing power management and control technologies for transportation engineering construction sites.

[0077] The relevant modules involved in this system are all hardware system modules or functional modules that combine computer software programs or protocols with hardware in the prior art. The computer software programs or protocols involved in these functional modules are technologies known to those skilled in the art and are not improvements to this system. The improvement of this system lies in the interaction or connection between the modules, that is, in improving the overall structure of the system to solve the corresponding technical problems that this system aims to address.

[0078] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent management and control system for electrical safety at traffic engineering construction sites, characterized in that: It includes a perception layer, a transmission layer, an edge computing layer, a cloud platform layer, and an application layer, with each layer interacting with data via the MQTT protocol; The sensing layer includes an electrical parameter acquisition module, an environmental parameter acquisition module, and an equipment status acquisition module. The electrical parameter acquisition module is deployed in various levels of distribution boxes and the power supply end of large construction machinery to collect current, voltage, leakage current, and cable joint temperature data. The environmental parameter acquisition module is used to collect ambient temperature, humidity, and dust concentration data. The equipment status module is used to collect the start-stop status of construction machinery and the open / closed status of distribution boxes. The transmission layer adopts a dual-mode networking structure of 4G / 5G main communication link and LoRa backup communication link, including an edge gateway with built-in breakpoint resume module and LoRa repeater. When the main link is interrupted, it automatically switches to the backup link and caches data locally. After the link is restored, the cached data is uploaded according to the timestamp. The edge computing layer is deployed in the computing chip of the edge gateway and has a built-in data preprocessing module and an environment-electrical parameter linkage early warning module, which are used to preprocess the data collected by the perception layer and realize local hierarchical early warning. The cloud platform layer is deployed on a cloud server and integrates an AI fault diagnosis model and a hidden danger data traceability module. It is used to receive data from the edge computing layer and realize full-link data storage for fault diagnosis and hidden danger handling. The application layer includes a monitoring screen, a mobile terminal APP, and a compliance report generation module, which are used to push early warning information, visualize electricity consumption data, and output compliance reports.

2. The intelligent control system for electrical safety at traffic engineering construction sites according to claim 1, characterized in that, The data preprocessing module of the edge computing layer performs outlier filtering and data normalization. Outlier filtering calculates the mean and standard deviation of electrical parameters over 10 consecutive acquisition cycles, and identifies and removes values ​​that deviate from the mean by 3 times the standard deviation. The formula for data normalization is: In the formula This is the original collected data. This is the lower limit for parameter monitoring. This represents the upper limit for parameter monitoring.

3. The intelligent control system for electrical safety at traffic engineering construction sites according to claim 1, characterized in that, The environmental-electrical parameter linkage early warning module of the edge computing layer has a built-in early warning coefficient calculation model, and the model formula is as follows: In the formula To provide a comprehensive early warning coefficient, This is the humidity weighting factor (value 0.4). This is the temperature weighting coefficient (value 0.3). This is the dust weighting coefficient (value 0.3). This represents the ambient humidity value. This refers to the temperature value of the cable connector. This refers to the dust concentration value. The module dynamically adjusts the electrical parameter safety threshold according to environmental conditions. When the ambient humidity is >85%RH, the leakage current warning threshold is lowered to 25mA. When the cable joint temperature is >80℃ and the dust concentration is >10mg / m³, a level 2 warning is triggered.

4. The intelligent control system for electrical safety at traffic engineering construction sites according to claim 1, characterized in that, The AI ​​fault diagnosis model of the cloud platform layer is a BP neural network model. In the model construction phase, the input layer is set with 5 nodes corresponding to normalized data of current, voltage, leakage current, cable temperature, and ambient humidity, respectively. The hidden layer has 2 layers with 12 and 8 nodes respectively. The output layer has 6 nodes corresponding to 6 types of electrical faults: short circuit, leakage, overload, over-temperature, voltage instability, and current fluctuation. In the model training phase, 10,000 sets of historical fault data are used, with cross-entropy as the loss function. The Adam optimizer is used to iteratively train the model at a learning rate of 0.001 until the convergence accuracy is 0.

001. The model completes self-iterative optimization with an iteration cycle of 30 days. After the model is deployed, it receives preprocessed data from the edge computing layer and outputs fault diagnosis results. The fault identification accuracy of the updated model is not less than 95%.

5. The intelligent control system for electrical safety at traffic engineering construction sites according to claim 1, characterized in that, The cloud platform layer's hazard data traceability module constructs a complete data link for hazard handling, including full-node data such as hazard discovery time, hazard type, responsible person for notification, handling time, handling measures, and review results. The module uses a hash encryption algorithm to encrypt the data; the encryption formula is... In the formula For encrypted digest, Original data for hazard mitigation. The encrypted data is stored in a distributed storage architecture in association with the original data, with a storage period of 5 years. Each data entry is associated with a unique electronic signature to achieve full-chain traceability.

6. The intelligent control system for electrical safety at traffic engineering construction sites according to claim 1, characterized in that, The electrical parameter acquisition module of the sensing layer has a current monitoring accuracy of no less than 0.5A, a voltage monitoring accuracy of no less than 0.1V, and a leakage current monitoring resolution of no more than 0.01mA, with an acquisition frequency of 1 second / time. The environmental parameter acquisition module has a temperature and humidity monitoring range covering -20℃ to 60℃, 10%RH to 95%RH, and a dust concentration monitoring resolution of no more than 0.01mg / m³, with an acquisition frequency of 5 seconds / time. The equipment status acquisition module is triggered by real-time acquisition of status changes, and the acquired data is encrypted using the national cryptographic SM4 encryption algorithm before transmission, with a transmission delay of no more than 100ms.

7. The intelligent control system for electrical safety at traffic engineering construction sites according to claim 1, characterized in that, The local cache capacity of the edge gateway in the transmission layer is no less than 16GB, and the communication coverage radius of the LoRa repeater is no less than 800 meters.

8. The intelligent control system for electrical safety at traffic engineering construction sites according to claim 1, characterized in that, The edge computing layer also includes a strategy pre-execution module, which pre-stores basic power control strategies such as overload power-off and timed power-off at night. When communication with the cloud platform layer is interrupted, the overload power-off strategy is automatically executed. The overload power-off trigger condition is that the circuit current exceeds 1.2 times the rated current for 30 seconds. The permission management module of the cloud platform layer is set with three levels of permissions. The first level of permission is that the project leader can view the power consumption data of the entire project and execute the global power-off operation. The second level of permission is that the safety officer can view the power consumption data of the area under his / her charge and execute the area power-off operation. The third level of permission is that the team leader can only view the power consumption data of the team. The authorization verification uses a dual verification method of account password plus dynamic verification code. The dynamic verification code is valid for 5 minutes.

9. The intelligent control system for electrical safety at traffic engineering construction sites according to claim 1, characterized in that, The mobile terminal APP of the application layer adopts a hierarchical warning delivery method. The first level of warning corresponds to short circuit and leakage faults, and is delivered through sound and light alarms, SMS and APP push. The second level of warning corresponds to overload and over-temperature faults, and is delivered through APP push and telephone reminder. The third level of warning corresponds to parameter fluctuation faults, and is delivered through APP push. The warning response time does not exceed 3 minutes. The compliance report generation module automatically generates daily / weekly / monthly electricity safety reports, and the report format complies with the filing requirements of the "Technical Specification for Temporary Electricity Safety at Construction Sites".

10. A method for intelligent management and control of electrical safety at traffic engineering construction sites, characterized in that: To implement the intelligent control system for electrical safety at traffic engineering construction sites as described in claims 1-9, the following steps are included: Step 1: The electrical parameter acquisition module, environmental parameter acquisition module, and equipment status acquisition module of the sensing layer acquire corresponding data respectively. After the acquisition is completed, the data is encrypted and transmitted to the transmission layer. Step 2: The transport layer transmits data through the 4G / 5G main link. When the main link is interrupted, it switches to the LoRa backup link. At the same time, the edge gateway caches the data. After the link is restored, the cached data is uploaded according to the timestamp. Step 3: The edge computing layer performs outlier filtering and normalization preprocessing on the received data, and then calculates the comprehensive early warning coefficient through the environment-electrical parameter linkage early warning module. Based on the coefficient, the corresponding level of local early warning is triggered. When cloud communication is interrupted, the preset power control strategy is executed. The environment-electrical parameter linkage early warning module automatically adjusts the leakage current early warning threshold according to the ambient humidity value. When the ambient humidity is >85%RH, the leakage current early warning threshold is lowered from 30mA to 25mA. When the cable joint temperature is >80℃ and the dust concentration is >10mg / m³, a level 2 early warning is triggered and the early warning data is uploaded to the cloud platform layer simultaneously. Step 4: The cloud platform layer receives preprocessed data from the edge computing layer, outputs the fault type through the AI ​​fault diagnosis model, and simultaneously the hidden danger data traceability module encrypts and stores the hidden danger handling data and generates an electronic signature; the self-iterative optimization process of the AI ​​fault diagnosis model is to collect 100 sets of new fault data every 30 days, supplement them to the training sample set and retrain the model, and the fault identification accuracy of the updated model is no less than 95%; Step 5: The mobile terminal APP at the application layer receives the graded early warning information, the monitoring screen displays the visualized content of electricity consumption data, and the compliance report generation module outputs the electricity safety report that meets the filing requirements.

Images