Anti-high-falling intelligent monitoring system for power transmission line operation

Abstract

The present application relates to the technical field of power transmission line operation safety monitoring, and discloses an intelligent monitoring system for preventing high-falling in power transmission line operation, which comprises a data acquisition end, an edge computing module, a cloud platform and an alarm response module; the data acquisition end constructs a multi-source perception network of "person-equipment-environment" by means of an intelligent safety helmet, a safety belt and auxiliary equipment, and accurately collects position, posture, stress and environmental parameters; the edge computing module preloads a dangerous action recognition model trained specially, and realizes real-time screening of abnormal data; the cloud platform deeply analyzes data in combination with a machine learning algorithm, and visualizes the operation scene by means of digital twin technology; and the alarm response module is provided with three thresholds, and realizes multi-end collaborative response and event whole-process tracing. The present application solves the problems of existing passive protection, single monitoring and lagging response, realizes accurate identification of high-falling risk, real-time early warning and collaborative disposal, and significantly improves the safety and management efficiency of power transmission line operation.

Description

Technical Field

[0001] This invention relates to the field of safety monitoring technology for power transmission line operations, and is a smart monitoring system for preventing falls from heights during power transmission line operations. Background Technology

[0002] In power transmission line operations, workers often need to perform installation, inspection, and maintenance at heights. Falls from height are the primary risk threatening the lives of these workers, often resulting in irreparable casualties and significant property damage. Therefore, fall protection has always been a core focus of safety management in power transmission line operations. Current technologies for fall protection largely rely on traditional passive protective equipment, such as ordinary safety helmets and safety belts. This type of equipment can only provide some buffering protection after an accident has occurred; it cannot proactively identify and warn of potential fall risks. Its proactive and forward-looking nature is severely lacking, making it difficult to prevent accidents from the outset.

[0003] With the development of intelligent technologies, simple safety monitoring equipment has begun to appear in some fields. However, in power transmission line operation scenarios, existing monitoring equipment has many shortcomings. On the one hand, the data collection dimensions are limited, mostly focusing only on the location information or single posture data of the workers, failing to comprehensively consider key influencing factors such as the stress state of the equipment and the working environment. This results in insufficient comprehensiveness and accuracy of risk identification, and is prone to false alarms or missed alarms. On the other hand, data processing relies heavily on centralized cloud computing. Power transmission line operation areas are often located in remote areas with unstable signals, resulting in high data transmission delays. This makes it impossible to achieve real-time risk analysis and rapid response, and it is difficult to provide sufficient time for workers and managers to handle emergencies.

[0004] Furthermore, existing monitoring systems suffer from poor adaptability in their risk identification models. Most are general-purpose models that haven't been specifically trained for the unique scenarios of power transmission line operations (such as different work types and complex environmental conditions). This makes it difficult to accurately distinguish between normal and dangerous actions, and the models lack iterative optimization capabilities, failing to adapt to constantly changing operational needs. Simultaneously, alarm response methods are relatively simplistic, often relying on single-device audible and visual alarms. This lack of multi-terminal collaborative response among operators, site managers, and the monitoring center leads to untimely alarm information transmission and inefficient handling processes, impacting emergency response efficiency. These problems make existing technologies insufficient to meet the actual needs of fall protection during power transmission line operations, necessitating a more comprehensive, accurate, and efficient intelligent monitoring system to address these issues. Summary of the Invention

[0005] This invention provides an intelligent monitoring system for preventing falls from heights during power transmission line operations, which overcomes the shortcomings of the existing technologies. It can effectively solve the problems of passive, single-dimensional monitoring, low accuracy of risk identification, and poor response coordination of existing fall protection methods.

[0006] The technical solution of the present invention is achieved through the following measures: an intelligent monitoring system for preventing falls from heights during power transmission line operations, comprising a data acquisition terminal, an edge computing module, a cloud platform, and an alarm response module.

[0007] Data acquisition terminal: including smart safety helmet, smart safety belt and auxiliary acquisition equipment. The smart safety helmet integrates positioning module and motion sensing module to collect the operator's location information, head movement status and body posture data; the smart safety belt integrates force sensing module to monitor the force status of safety equipment; the auxiliary acquisition equipment is used to collect environmental parameters that affect the risk of falling from height, such as wind speed, temperature and light. Edge computing module: Preloaded with a machine learning-based dangerous action recognition model, it performs real-time preliminary analysis of multi-source data collected from the acquisition end, filters abnormal data and transmits it to the cloud platform to achieve low-latency risk identification; the edge computing module and the cloud platform transmit full data through fiber optic or 5G private network to ensure data transmission stability; the edge computing module and the alarm response module achieve millisecond-level real-time communication through Bluetooth Low Energy to ensure zero-delay warning response.

[0008] Cloud platform: Used to store full operational data and abnormal data, and to achieve risk assessment and trend prediction through big data analysis.

[0009] Alarm Response Module: Based on the movement status of the operator, the stress status of the equipment, and environmental parameters, multi-level alarm thresholds are set. Based on the collaborative judgment results of the edge computing module and the cloud platform, it triggers multi-channel responses to the operator, on-site management personnel, and monitoring center, and records the alarm time, location, alarm level, and handling measures.

[0010] The following are further optimizations and / or improvements to the above-mentioned technical solution: The aforementioned data acquisition terminal, edge computing module, and cloud platform can be connected in sequence, and the edge computing module and alarm response module can be connected in sequence.

[0011] The aforementioned data acquisition terminals may include smart safety helmets, smart safety belts, and auxiliary acquisition devices. The smart safety helmets collect the location information and head movement data of the workers through positioning modules and motion sensing modules; the smart safety belts monitor their own force state and the workers' body posture data through force sensing modules and posture sensing modules; and the auxiliary acquisition devices collect working environment parameters that affect the risk of falls through heights through environmental sensing modules.

[0012] The aforementioned positioning module can be a BeiDou positioning module, which combines multi-base station differential technology to achieve high-precision positioning. The motion sensing module may include a three-axis accelerometer and a gyroscope. The force sensing module may be a strain sensor, and the attitude sensing module may be a tilt sensor. The auxiliary acquisition equipment may be deployed at the crossarm or ladder entrance of the transmission tower.

[0013] The aforementioned data acquisition terminal can transmit the acquired data to the edge computing module via a multi-mode fused wireless communication link.

[0014] The aforementioned dangerous action recognition model can be a deep learning-based model trained on a large-scale power transmission line operation dataset. The dataset covers personnel action data under different operation types and environmental conditions. The model can distinguish between normal operation actions and dangerous actions, and can continuously incorporate new operation scenario data to achieve adaptive iterative optimization.

[0015] The aforementioned cloud platform can utilize machine learning algorithms for in-depth data analysis, including random forest and LSTM neural network algorithms.

[0016] The aforementioned multi-level alarm thresholds may include a first-level warning threshold, a second-level warning threshold, and a third-level alarm threshold. The first-level warning threshold corresponds to a body tilt angle exceeding the normal working range or abnormal fluctuations in seat belt tension. The second-level warning threshold corresponds to an abnormal vertical acceleration that continues to reach a preset duration threshold. The third-level alarm threshold corresponds to the detection of an instantaneous and severe vertical acceleration impact.

[0017] The aforementioned multi-channel response may include sending audible and visual alarms to the smart devices worn by workers, sending pop-up and voice alarms to the mobile apps of on-site managers, and pushing alarm information to the monitoring center and linking it to GIS map positioning.

[0018] The aforementioned cloud platform can recreate work scenarios using digital twin technology, intuitively presenting personnel locations, equipment status, and risk distribution.

[0019] This invention constructs a comprehensive, end-to-end fall risk monitoring and protection system through multi-module collaborative operation. The data acquisition terminal collects key data from multiple dimensions, including personnel, equipment, and environment, ensuring no risk factors are overlooked. The edge computing module enables rapid data processing locally, filtering abnormal data before transmitting it to the cloud platform, effectively reducing transmission latency and ensuring real-time response. The hazardous action recognition model is specially trained and adaptively iterative, significantly improving the accuracy of risk identification. The combination of multi-level alarm thresholds and multi-channel response mechanisms enables tiered risk warning and collaborative handling, ensuring timely delivery of alarm information to relevant personnel. The cloud platform's deep analysis and digital twin technology application provide strong support for risk assessment, prediction, and operation management. This invention changes the traditional passive protection model, achieving proactive identification, real-time warning, rapid response, and scientific handling of fall risks, significantly improving the safety of power transmission line operations. It also optimizes operation management processes, improves management efficiency, and provides comprehensive and reliable technical protection for the safety of power transmission line operations. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the intelligent monitoring system for preventing falls from heights during power transmission line operations, according to an embodiment of the present invention.

[0021] Figure 2 This is a schematic diagram of the training process of the dangerous action recognition model according to an embodiment of the present invention.

[0022] Figure 3 This is a schematic diagram of the training process of the random forest algorithm according to an embodiment of the present invention.

[0023] Figure 4 This is a schematic diagram of the training process of the LSTM neural network algorithm according to an embodiment of the present invention. Detailed Implementation

[0024] The present invention is not limited to the following embodiments, and the specific implementation can be determined according to the technical solution of the present invention and the actual situation.

[0025] The present invention will be further described below with reference to embodiments: Example 1: As Figure 1As shown, this embodiment provides an intelligent monitoring system for preventing falls from heights during power transmission line operations. It includes a data acquisition terminal, an edge computing module, a cloud platform, and an alarm response module. The data acquisition terminal collects personnel location information, head movement data, body posture data, monitors equipment stress data, and collects environmental parameters affecting fall risk. The edge computing module pre-loads a hazardous action recognition model to perform preliminary analysis of the collected data, filter abnormal data, and transmit the abnormal data to the cloud platform. The cloud platform stores the operational data collected by the data acquisition terminal and the abnormal data transmitted by the edge computing module, and performs in-depth analysis to achieve risk assessment and prediction. The alarm response module sets multi-level alarm thresholds based on personnel movement, equipment stress, and environmental parameters. Based on the collaborative judgment results of the edge computing module and the cloud platform, it triggers multi-channel responses to personnel, on-site managers, and the monitoring center, and records the alarm time, location, alarm level, and response measures. This system achieves intelligent fall risk monitoring through a collaborative process of "data acquisition - edge processing - cloud analysis - multi-terminal response." During operation, the data acquisition terminal simultaneously collects information from three dimensions: personnel, equipment, and environment. The smart safety helmet captures the worker's position and head movement, the smart safety belt monitors the force exerted on the worker and their body posture, and auxiliary equipment collects environmental parameters affecting fall risks. The collected data is transmitted to the edge computing module via a multi-mode fusion wireless link. The module performs preliminary analysis in real time using a pre-loaded hazardous action recognition model, quickly filtering abnormal data and pushing it to the cloud platform to avoid delays caused by invalid data transmission. The cloud platform stores all data and abnormal data, using machine learning algorithms to deeply mine risk patterns and achieve risk assessment and prediction. The alarm response module combines real-time judgment from edge computing with in-depth analysis from the cloud, comparing the data against multi-level alarm thresholds set based on personnel, equipment, and environmental parameters. It triggers multi-channel responses, including sound and light, APP pop-ups, and GIS-linked positioning, to the workers, site managers, and monitoring center, simultaneously recording key alarm information. This forms a closed-loop working mechanism of "real-time monitoring - accurate identification - tiered early warning - collaborative response," effectively preventing fall risks. This approach enables comprehensive collection and collaborative processing of multi-dimensional data, capturing potential risks of falls from heights at their source. By combining edge computing with cloud analytics, the real-time and accurate identification of risks is ensured, while multi-level alarms and multi-channel responses guarantee timely and coordinated risk management, effectively improving the level of operational safety.

[0026] In this embodiment, the data acquisition terminal, edge computing module, and cloud platform are sequentially connected for communication. The edge computing module is also connected for communication with the alarm response module. The communication connection adopts a stable and reliable communication protocol to ensure efficient and accurate data transmission between modules. This allows for the construction of a complete data transmission link, ensuring smooth connection of the entire process from data acquisition, processing, analysis to response. It avoids problems such as missed risk detection and untimely response caused by communication interruption or delay between modules, thereby improving the overall stability and reliability of the system.

[0027] In this embodiment, the data acquisition terminal includes a smart safety helmet, a smart safety belt, and auxiliary acquisition equipment. The smart safety helmet collects the worker's location information and head movement data through a positioning module and a motion sensing module. The smart safety belt monitors its own force state and the worker's body posture data through a force sensing module and a posture sensing module. The auxiliary acquisition equipment collects working environment parameters that affect the risk of falls from heights through an environmental sensing module. The smart safety helmet and smart safety belt are worn by the worker, and the auxiliary acquisition equipment is fixed at a key position in the working area. In this way, comprehensive data coverage of the worker's status, the status of protective equipment, and the working environment can be achieved, ensuring that no key factors that may cause a fall from height accident are missed, providing comprehensive and sufficient data support for risk identification, and improving the comprehensiveness and accuracy of risk identification.

[0028] In this embodiment, the positioning module is a BeiDou positioning module, which combines multi-base station differential technology to achieve high-precision positioning. The motion sensing module includes a three-axis accelerometer and a gyroscope; the force sensing module is a strain sensor; and the attitude sensing module is a tilt sensor. The auxiliary acquisition equipment is deployed at the crossarm or ladder entrance of the transmission tower. The BeiDou positioning module can accurately obtain the real-time position of the workers. The three-axis accelerometer and gyroscope can capture data such as the acceleration and angular velocity of head movements. The strain sensor can sense the changes in the tension of the safety belt, and the tilt sensor can detect the tilt angle of the workers' bodies. This improves the accuracy of data acquisition, ensuring that the data in each dimension can truly reflect the actual situation and provide high-quality data input for the risk identification model. At the same time, the auxiliary acquisition equipment is deployed in key locations to collect the most representative environmental parameters, further improving the accuracy of risk assessment.

[0029] In this embodiment, the data acquisition terminal transmits the acquired data to the edge computing module through a multi-mode fusion wireless communication link. The multi-mode fusion wireless communication link can automatically switch communication modes according to the signal strength of the working area. This can adapt to the complex communication environment of the power transmission line working area, ensure the stability and continuity of data transmission, avoid data loss or transmission delay caused by poor signal in a single communication mode, and ensure that the edge computing module can acquire and process the acquired data in a timely manner.

[0030] In this embodiment, the hazardous action recognition model is a deep learning-based model trained on a large-scale power transmission line operation dataset. This dataset covers personnel action data under different operation types and environmental conditions. The model can distinguish between normal and hazardous actions and continuously incorporates new operation scenario data for adaptive iterative optimization. During training, the model learns a large number of typical action features of power transmission line operations, enabling it to accurately identify hazardous actions with a high risk of fall. This improves the accuracy of risk identification, reduces false alarms and missed alarms, and the model's adaptive iterative capability allows the system to adapt to constantly changing operation scenarios and requirements, extending the system's lifespan and applicability. Figure 2 As shown, the dangerous action recognition model in this invention is a deep learning model based on a CNN-LSTM hybrid neural network. Its internal structure includes a feature extraction layer, a temporal fusion layer, and a classification output layer: the feature extraction layer uses a 3-layer convolutional neural network (CNN) to extract spatial features of the collected data through 16, 32, and 64 convolutional kernels in successive increments; the temporal fusion layer uses a 2-layer long short-term memory network (LSTM) to capture the temporal dependencies of the data; the classification output layer uses a fully connected layer and a softmax activation function to output the classification results of normal or dangerous actions. Each layer is connected to a dropout layer (with a dropout rate set to 0.2) through a ReLU activation function to avoid overfitting. During model training, over 100,000 power transmission line operation sample data were first collected, covering normal actions such as climbing and tool passing, as well as dangerous actions such as falls, imbalances, and violations. Each sample included a time-series sequence of position, motion state, force data, and environmental parameters (sampling frequency 100Hz, single sample duration 3 seconds), and was labeled with the binary tags "normal" and "dangerous". The dataset was then divided into training, validation, and test sets in a 7:2:1 ratio. The Adam optimizer was used (initial learning rate 0.001, decaying by 10% every 50 rounds), with the cross-entropy loss function as the optimization objective. Iterative training was conducted for 200 rounds until the validation set accuracy stabilized above 95%. During training, an early stopping strategy (stopping if the validation set loss did not decrease for 10 consecutive rounds) was used to prevent overfitting, ultimately resulting in the trained model. When the model is applied, the input data is multi-dimensional time-series data collected in real time by the data acquisition terminal. After standardization processing (mapping the data to the [0,1] interval), it is input into the model. The feature extraction layer extracts spatial features from the data of each dimension. The time-series fusion layer integrates the feature information of different time steps. The classification output layer outputs the action category and confidence level. When the confidence level is higher than 90% and it is determined to be a dangerous action, it is determined to be abnormal data and transmitted to the cloud platform.

[0031] In this embodiment, the cloud platform utilizes machine learning algorithms for deep data analysis. These algorithms include random forest and LSTM neural network algorithms. Random forest algorithms can process multi-dimensional data and perform accurate classification and regression analysis, while LSTM neural network algorithms excel at capturing dependencies in time-series data. This allows for in-depth mining of massive amounts of collected data, extracting potential risk patterns and trends, and achieving accurate assessment and prediction of fall risks. This provides a scientific basis for operational management decisions, enabling proactive and targeted protective measures. Figure 3 As shown, in this invention, the Random Forest algorithm focuses on ranking the importance of multi-dimensional risk factors. Its input data includes all raw data collected by the data acquisition terminal (covering worker position accuracy data, head motion acceleration and angular velocity data, body tilt angle data, seatbelt tension change data, and environmental parameters such as wind speed, wind direction, and temperature) and abnormal data output by the edge computing module (including dangerous action identification results, data anomaly annotations, and preliminary risk type assessments). During algorithm construction, the number of decision trees is set to 100, with a maximum depth of 10 layers per tree. The Gini coefficient is used as the feature splitting criterion. The bootstrap sampling method randomly selects 80% of the samples from the input data to construct the training set for each decision tree, with the remaining 20% ​​used for internal validation. During training, the algorithm statistically analyzes the splitting contribution of each risk factor across all decision trees, ultimately outputting the importance weight of each risk factor (weight values ​​range from 0 to 1, with higher weights indicating greater impact on fall risk), providing data support for subsequent risk warning threshold optimization and key protection direction determination. Figure 4 As shown, the LSTM neural network algorithm focuses on risk trend prediction based on historical data. The input data is consistent with the random forest algorithm, and additionally includes historical operation data from the past 3 months (summarized daily, including daily frequency of abnormal data, number of risk occurrences, and operation duration under different environmental conditions). The algorithm structure is set to 3 layers of LSTM hidden layers, with 64, 32, and 16 neurons in each layer, respectively. The input sequence length is set to 7 days (i.e., predicting subsequent risks based on multi-dimensional data from 7 consecutive days). The output layer uses a fully connected layer and a Sigmoid activation function to output the risk prediction results and corresponding probabilities for high, medium, and low levels for the next 24 hours. During training, the Adam optimizer is used, with an initial learning rate of 0.0005, decaying by 50% every 30 iterations, using mean squared error as the loss function, and iterating for 300 epochs. The prediction accuracy is monitored in real time using a validation set, and training stops when the prediction accuracy is consistently above 92%. By capturing the temporal dependencies in the data, this algorithm can effectively predict the changing trends of fall risk under different working periods and environmental conditions, helping managers to formulate targeted protective measures in advance and further improve the system's forward-looking and effective risk prevention and control.

[0032] In this embodiment, the multi-level alarm thresholds include a first-level warning threshold, a second-level warning threshold, and a third-level alarm threshold. The first-level warning threshold corresponds to a body tilt angle exceeding the normal working range or abnormal fluctuations in seat belt tension. The second-level warning threshold corresponds to an abnormal vertical acceleration that persists for a preset duration threshold. The third-level alarm threshold corresponds to the detection of a sudden and severe vertical acceleration impact. Different levels of thresholds correspond to different levels of risk. This allows for graded handling of risks, enabling corresponding response measures to be taken for different risk levels. This avoids overreactions caused by minor risks while ensuring that serious risks can be dealt with promptly and effectively, thus improving the scientific and rational nature of emergency response.

[0033] In this embodiment, the multi-channel response includes sending audible and visual alarms to the smart devices worn by workers, sending pop-up and voice alarms to the mobile APP of on-site managers, and pushing alarm information to the monitoring center and linking it with GIS map positioning. The audible and visual alarms can directly remind workers to take timely avoidance measures, while the pop-up and voice alarms can enable on-site managers to quickly know the risk situation and go to deal with it. The monitoring center can grasp the overall risk status in real time and make overall arrangements. In this way, alarm information can be transmitted synchronously from multiple terminals, ensuring that relevant personnel can obtain risk information and carry out collaborative handling as soon as possible, shortening emergency response time, improving handling efficiency, and minimizing accident losses.

[0034] In this embodiment, the cloud platform uses digital twin technology to recreate the work scenario, intuitively presenting the personnel location, equipment status, and risk distribution. The virtual scenario constructed by digital twin technology is synchronized with the actual work scenario in real time. This allows managers to clearly grasp the overall situation of the work site remotely, promptly identify potential risks, and conduct dispatch and command. It also provides an intuitive basis for accident review, which helps to optimize subsequent protective measures and work processes. In this invention, the digital twin technology applied to the cloud platform constructs a virtual scene that precisely matches the actual operation scenario through the logic of "data modeling - real-time mapping - dynamic updating". The core process is as follows: First, based on the CAD design drawings of physical facilities such as transmission line towers, ladders, and crossarms, and on-site laser scanning data, a 1:1 scale three-dimensional geometric model is constructed, and basic attribute data such as equipment parameters and geographical information of the operation area are entered at the same time. Second, through the data interface, the location of personnel, the stress state of equipment, environmental parameters, and the risk assessment results of the edge computing module are accessed in real time from the data acquisition terminal. The dynamic data of the physical entity is associated with the corresponding dimensions of the virtual model to achieve real-time mapping of personnel movement trajectory, equipment status changes, and risk area distribution. Finally, a data synchronization update mechanism is established with a synchronization period of 50ms. Through a multi-mode communication link, dynamic data of the physical scene is continuously received, and the state parameters of the virtual model are automatically corrected to ensure that the virtual scene and the actual operation scenario are completely synchronized in terms of spatial location, equipment status, and risk distribution. Managers can intuitively view the real-time location of workers, the operating status of safety equipment, and the distribution of high-risk areas through virtual scenarios, providing visual support for remote dispatching, emergency response, and operation planning, and improving the intuitiveness and efficiency of operation management.

[0035] In this invention, the data transmission link adopts a "wireless multi-mode fusion" transmission scheme, integrating 4G / 5G, Wi-Fi, and LoRa technologies. It features adaptive switching and AES-256 data encryption. In 5G mode, the transmission latency is controlled within 50ms, and in LoRa mode, the transmission distance can reach several kilometers. Wi-Fi is used for rapid data synchronization in local areas, ensuring secure and stable data transmission in different operational scenarios. Edge computing nodes are deployed on maintenance vehicles, smart micro-stations, or pole-mounted auxiliary facilities near the operational area, equipped with lightweight AI inference chips. The cloud platform uses a distributed database to store massive amounts of operational data, supporting historical data backtracking and multi-dimensional queries. The virtual scene constructed through digital twin technology is synchronized in real time with the actual operational scene, providing managers with visual decision support. Specific parameters are set for multi-level alarm thresholds: Level 1 alarm corresponds to a body tilt angle exceeding ±30° or abnormal fluctuations in seatbelt tension; Level 2 alarm corresponds to a vertical acceleration >8m / s² and a duration >0.2s; Level 3 alarm corresponds to a vertical acceleration >15m / s². The edge computing nodes and the cloud platform collaboratively determine the alarm level. In the multi-channel response, the smart devices worn by the workers include smart safety helmets and smart bracelets. On-site managers receive pop-up and voice alarms via a mobile app. The monitoring center, in conjunction with GIS map positioning, initiates the emergency response process. The system automatically records alarm-related information for later review. Auxiliary data acquisition equipment can be configured with visual sensors to help identify the standardization of workers' movements and information about obstacles around the equipment. In some scenarios, it can also be combined with drones for collaborative operations. A linkage module can be developed to avoid interference risks. During nighttime operations, the combination of visual sensors and infrared imaging technology improves the accuracy of movement recognition.

[0036] During operation, the smart safety helmet, smart safety belt, and auxiliary data acquisition equipment at the data acquisition end are activated simultaneously to collect data such as the worker's position, head movement, body posture, equipment stress state, and environmental parameters. This data is transmitted to the edge computing module via a multi-mode fusion wireless communication link. The edge computing module uses a pre-loaded hazardous action recognition model to perform real-time analysis of the collected data, filtering out abnormal risk data and transmitting it to the cloud platform. The cloud platform stores all collected and abnormal data, performs in-depth analysis using machine learning algorithms, and combines historical and real-time data to achieve risk assessment and prediction. The alarm response module, based on the collaborative judgment results of the edge computing module and the cloud platform, compares the results with multi-level alarm thresholds, triggers corresponding multi-channel alarm responses, and records alarm-related information. Management personnel can monitor the operation in real-time through the digital twin scenario on the cloud platform and coordinate emergency response efforts. The entire system achieves real-time monitoring, accurate identification, tiered early warning, and collaborative handling of fall risks, significantly improving the safety and management efficiency of power transmission line operations.

[0037] It should be noted that, in this invention, the edge computing module refers to a computing module deployed near the work area that can process collected data locally. Its core function is to reduce data transmission latency and achieve real-time data processing. The dangerous action recognition model refers to a model based on deep learning algorithms, trained on a large amount of power transmission line operation data, capable of distinguishing between normal and dangerous actions. Digital twin technology refers to constructing a virtual scene that completely maps to the physical work scene through digital means, enabling real-time monitoring and simulation analysis of the physical scene. The multi-mode fusion wireless communication link refers to a link that integrates multiple wireless communication technologies and can automatically switch communication modes according to the strength of environmental signals. The triaxial accelerometer refers to a sensor capable of measuring acceleration in three orthogonal directions in space, capturing changes in the acceleration of an object. The strain sensor refers to a sensor capable of sensing the strain changes produced by an object under force and converting it into an electrical signal. The positioning module uses a BeiDou-3 positioning module supporting multi-base station differential technology (positioning accuracy ≤1m); the motion sensing module uses a three-axis accelerometer with a sampling frequency ≥100Hz (measurement range ±16g) and a gyroscope (measurement range ±2000° / s); the force sensing module uses a strain sensor with a range of 0-50kN and an accuracy class of 0.1%FS; the attitude sensing module uses a tilt sensor with a measurement range of ±90° and an accuracy of ±0.1°; the environmental sensing module includes a wind speed sensor with a range of 0-25m / s, a wind direction sensor with a range of 0-360°, a temperature sensor with a range of -40℃ to 85℃, and a humidity sensor with a range of 0-100%RH; the multi-mode fusion wireless communication link supports 4G / 5G (downlink rate ≥100Mbps) and Wi-Fi. 6 (transmission rate ≥ 1.2Gbps) and LoRa (communication distance ≥ 3km) technologies; the edge computing module is equipped with a lightweight AI inference chip with a computing power ≥ 1TOPS; the cloud platform adopts a distributed database (supporting concurrent access ≥ 1000QPS), and its deployed random forest algorithm is set with 100 decision trees and a maximum depth of 10 layers per tree, and the LSTM neural network algorithm is set with 3 hidden layers (the number of neurons are 64, 32 and 16 respectively) and the input sequence length is 7 days. The smart safety helmet is an industrial-grade smart safety helmet that conforms to the GB 2811-2019 standard. The shell is made of ABS engineering plastic, and its impact resistance is sufficient to withstand a 1kg steel ball falling freely from a height of 1m without breaking. The built-in positioning module is a Beidou-3 B1C / B2a dual-frequency module (positioning accuracy ≤0.5m). The motion sensing module integrates a three-axis accelerometer with a sampling frequency of 100Hz (measurement range ±16g) and a gyroscope (measurement range ±2000° / s). It is equipped with a lithium polymer battery (capacity 3000mAh) with a battery life of ≥12 hours, supports IP67 dust and water resistance, and integrates an audible and visual alarm unit (alarm volume ≥85dB, alarm light visible distance ≥50m) and a Bluetooth 5.0 communication module.The smart safety belt complies with GB 6095-2021 standards. The webbing is made of high-strength polyester fiber (tensile strength ≥22kN), and it is equipped with 2 main attachment points and 1 auxiliary attachment point. The built-in force sensing module is a strain sensor with a range of 0-50kN and an accuracy of 0.1%FS (sampling frequency 50Hz), and the attitude sensing module is an tilt sensor with a measurement range of ±90° and an accuracy of ±0.1°. It integrates a lithium polymer battery (capacity 5000mAh) with a battery life of ≥24 hours, supports IP65 dust and water resistance, transmits data in real time through a LoRa communication module (communication distance ≥3km), and has a built-in low battery alarm function.

[0038] The above technical features constitute the embodiments of the present invention, which have strong adaptability and implementation effect. Unnecessary technical features can be added or removed according to actual needs to meet the needs of different situations.

Claims

1. A smart monitoring system for preventing falls from heights during power transmission line operations, characterized in that, Includes data acquisition terminal, edge computing module, cloud platform and alarm response module; The data acquisition terminal is used to collect the location information of workers, head movement data, body posture data, monitor the stress state data of equipment, and collect the working environment parameters that affect the risk of falls from heights; The edge computing module preloads a dangerous action recognition model to perform preliminary analysis of the collected data and filter out abnormal data, and transmits the abnormal data to the cloud platform. The cloud platform is used to store operational data collected by the data acquisition terminal and abnormal data transmitted by the edge computing module, and to perform in-depth analysis of the data to achieve risk assessment and prediction. The alarm response module sets multi-level alarm thresholds based on the movement status of the operator, the stress status of the equipment, and environmental parameters. It is used to trigger multi-channel responses to the operator, on-site management personnel, and monitoring center based on the collaborative judgment results of the edge computing module and the cloud platform, and to record the alarm time, location, alarm level, and handling measures.

2. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 1, characterized in that, The data acquisition terminal, edge computing module, and cloud platform are connected in sequence, and the edge computing module is connected in communication with the alarm response module.

3. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 1 or 2, characterized in that, The data acquisition terminal includes a smart safety helmet, a smart safety belt, and auxiliary acquisition equipment. The smart safety helmet collects the operator's location information and head movement data through a positioning module and a motion sensing module; the smart safety belt monitors its own force state and the operator's body posture data through a force sensing module and a posture sensing module; and the auxiliary acquisition equipment collects working environment parameters that affect the risk of falls through heights through an environmental sensing module.

4. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 3, characterized in that, The positioning module is a Beidou positioning module, which combines multi-base station differential technology to achieve high-precision positioning. The motion sensing module includes a three-axis accelerometer and a gyroscope. The force sensing module is a strain sensor, and the attitude sensing module is a tilt sensor. The auxiliary acquisition equipment is deployed at the crossarm or ladder entrance of the transmission tower.

5. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 1, 2, or 4, characterized in that, The data acquisition terminal transmits the acquired data to the edge computing module via a multi-mode fused wireless communication link.

6. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 1, 2, or 4, characterized in that, The dangerous action recognition model is a deep learning-based model trained on a large-scale power transmission line operation dataset. The dataset covers personnel action data under different operation types and environmental conditions. The model can distinguish between normal operation actions and dangerous actions, and can continuously incorporate new operation scenario data to achieve adaptive iterative optimization.

7. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 1, 2, or 4, characterized in that, The cloud platform uses machine learning algorithms for in-depth data analysis, including random forest and LSTM neural network algorithms.

8. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 1, 2, or 4, characterized in that, The multi-level alarm thresholds include a first-level warning threshold, a second-level warning threshold, and a third-level alarm threshold. The first-level warning threshold corresponds to the body tilt angle exceeding the normal working range or abnormal fluctuations in the tension of the safety belt. The second-level warning threshold corresponds to abnormal vertical acceleration that continues to reach a preset duration threshold. The third-level alarm threshold corresponds to the detection of instantaneous and severe vertical acceleration impact.

9. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 1, 2, or 4, characterized in that, The multi-channel response includes sending audible and visual alarms to the smart devices worn by workers, sending pop-up and voice alarms to the mobile APP of on-site managers, and pushing alarm information to the monitoring center and linking it to GIS map positioning.

10. The intelligent monitoring system for preventing falls from heights during power transmission line operations according to claim 1, 2, or 4, characterized in that, The cloud platform uses digital twin technology to recreate the work scenario, intuitively presenting the location of personnel, equipment status, and risk distribution.

Images