Telecommunications construction management system

Abstract

The present invention relates to a worker safety management system, and more specifically, to a telecommunications construction management system that collects workers' biosignals, behavioral patterns, and physical risk factors at the site in real time by reflecting the specific characteristics of a telecommunications construction site, precisely determines the overall risk level of workers and the site through hierarchical and dynamic logical rules based on the frequency and severity of risk events beyond simple threshold comparisons, and performs multiple notifications, including automated equipment control, when a risk situation occurs, thereby preemptively preventing safety accidents that may occur during the telecommunications construction process.

Description

Technology Field

[0001] The present invention relates to a worker safety management system, and more specifically, to a telecommunications construction management system that collects workers' biosignals, behavioral patterns, and physical risk factors at the site in real time by reflecting the specific characteristics of a telecommunications construction site, precisely determines the overall risk level of workers and the site through hierarchical and dynamic logical rules based on the frequency and severity of risk events beyond simple threshold comparisons, and performs multiple notifications, including automated equipment control, when a risk situation occurs, thereby preemptively preventing safety accidents that may occur during the telecommunications construction process. Background Technology

[0002] Recently, various safety management systems utilizing Internet of Things (IoT) sensors or video analysis technologies have been proposed to prevent safety accidents in industrial sites. These conventional systems have primarily used methods such as measuring workers' vital signs (heart rate, body temperature, etc.) to send warnings when set thresholds are exceeded, or detecting obvious rule violations, such as not wearing a safety helmet, through video analysis.

[0003] However, these conventional technologies have several limitations.

[0004] First, most systems have limitations in comprehensively evaluating the severity of risk because they only determine whether a single data point momentarily exceeds a threshold. For example, there was a problem in that they could not distinguish and judge much more fatal risky behaviors, such as unbuckling a seatbelt while working at heights, even if the heart rate is within the normal range.

[0005] Second, since it was developed for general industrial sites such as construction and manufacturing, it fails to specifically reflect the unique risk factors of telecommunications construction sites, such as climbing communication poles, working inside manholes, and working near live lines. Consequently, it lacked the capability to evaluate physical and potential on-site risk factors, such as the maintenance status of communication poles or open manholes, in conjunction with worker behavior.

[0006] Third, information transmission methods were often limited to passive measures, such as sounding an alarm or notifying a manager when a dangerous situation occurred. This led to the problem of failing to effectively prevent accidents when workers ignored the warnings or when managers were unable to respond immediately.

[0007] Accordingly, there is a need for a new type of safety management system that reflects the unique characteristics of telecommunications construction sites, organically combines various risk factors to assess the severity of risks in a three-dimensional manner, and can physically block risky behaviors immediately before an accident occurs. The problem to be solved

[0008] The present invention was devised to solve the problems of the prior art as described above, and its purpose is as follows.

[0009] The first task is to provide a telecommunications construction management system capable of comprehensively and precisely determining the severity of risk situations by going beyond merely judging whether instantaneous measurements exceed a standard, quantitatively counting the 'number' of occurrences of risk events, and providing a hierarchical judgment logic that prioritizes judgments based on the type of risk.

[0010] The second task is to provide a telecommunications construction management system capable of evaluating the comprehensive risk status of the entire site by independently assessing not only worker behavior but also site-specific physical and environmental risk factors, such as the number of uninspected poles or open manholes, alongside human risk factors, and organically combining them.

[0011] The third task is to provide a telecommunications construction management system capable of proactively preventing accidents without relying on human reaction, by going beyond simple warnings when dangerous situations occur and performing a physical interlock function that automatically blocks the operation of aerial work platforms or winches used by the relevant worker when specific behaviors directly linked to the risk of falling are detected.

[0012] The problems of the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0014] In a communication construction management system according to an embodiment of the present invention for solving the above problem,

[0015] A data collection unit that collects safety-related raw data from workers and the site;

[0016] A storage unit that stores preset risk conditions including a judgment rule table;

[0017] A control unit that receives raw data from the data collection unit, determines a risk level based on the judgment rule table stored in the storage unit, and generates a control signal according to the judgment result; and

[0018] A notification unit that transmits a warning to an operator or manager according to the control signal of the control unit; is included,

[0019] The above control unit is,

[0020] A data analysis module that processes the above raw data to generate quantitative count data indicating the number of occurrences of risk events; and

[0021] A risk judgment module that receives the quantitative count data from the data analysis module and determines the final risk level by applying the judgment rule table stored in the storage unit;

[0022] The above data analysis module is,

[0023] Based on raw heart rate data received from the biosignal detection unit of the above data collection unit, the number of heart rate surge events is generated by counting cases where the heart rate increases above a preset threshold value within a preset time, and

[0024] Based on data received from the position, attitude, and motion detection unit of the above data collection unit, the number of times the seat belt is disconnected is generated by counting the state in which the seat belt is disconnected for a preset time or longer within an elevated work zone at a height of 2 meters or more from the ground.

[0025] Based on image data received from the image capturing unit of the data collection unit, the number of episodes of not wearing a safety helmet is generated by counting cases where a worker is not wearing a safety helmet for a preset time or longer as one episode.

[0026] The above risk judgment module is used to determine the risk level of individual workers,

[0027] A first rule that immediately determines the risk level of the worker as 'danger' regardless of the value of other count data when the number of times the seatbelt is disconnected is one or more; and

[0028] A hierarchical judgment structure is characterized by performing a second rule that determines the risk level of the worker as 'Warning' or 'Caution' by comparing the sum of the number of heart rate surge events and the number of episodes of not wearing a safety helmet with a first threshold value stored in the storage unit when the above first rule does not apply.

[0029] The above risk assessment module is designed to evaluate the overall risk status at the site level,

[0030] The number of uninspected communication poles for which a preset period has elapsed since the last safety inspection date is obtained from the facility management database stored in the above storage unit, and the number of manholes with open lids is obtained from IoT sensors installed at the site to evaluate the physical risk level of the site,

[0031] It is characterized by evaluating the human risk level at the site by aggregating the number of high-risk workers determined to be 'warning' or 'danger' based on the judgment result of Paragraph 3 above and the total number of rule violation events recorded in the system,

[0032] The above risk judgment module is,

[0033] A step of independently evaluating the physical risk level and the human risk level as 'high', 'medium', and 'low', respectively, according to judgment rules stored in the storage unit; and

[0034] It is characterized by applying a combination rule that determines the overall risk status of the site as 'risk' regardless of the evaluation result of the other when either the physical risk level or the human risk level evaluated above is evaluated as 'high'.

[0035] The above control unit is,

[0036] If the above risk determination module determines the risk level of the worker as 'risk' according to the above first rule,

[0037] In addition to controlling the above notification unit to perform multiple notifications that simultaneously transmit voice and vibration warnings to the worker and send an emergency notification to the manager,

[0038] It may be characterized by the effect of preemptively preventing fall accidents by generating and transmitting an interlock signal that remotely blocks the operation of the aerial work vehicle or winch being used by the worker in whom the above-mentioned danger is detected, thereby physically forcibly stopping the continuation of dangerous behavior without relying on human reaction speed. Effects of the invention

[0040] According to the present invention for solving the above-mentioned problems, the following effects can be obtained.

[0041] It improves the accuracy and reliability of risk assessment. This invention adopts a hierarchical judgment structure that prioritizes judgments based on the severity of risky behaviors. For example, by identifying a single act of 'unbuckling the seatbelt' as the highest risk even if all other vital signs are normal, it is possible to proactively respond without missing fatal risk factors with the highest probability of an accident.

[0042] This enables comprehensive safety management specialized for telecommunications construction sites. The invention independently evaluates and combines not only worker behavior (human risk) but also the number of uninspected telecommunications poles or open manholes (physical risk). Through this, it is possible to identify and manage potential and environmental risk factors at the site that conventional worker-centric systems could not detect, thereby establishing a more comprehensive safety network.

[0043] It maximizes the practical effect of accident prevention. When a clear and fatal danger is detected, such as dangerous behavior immediately before a fall, this invention physically and forcibly stops the operation of aerial equipment simultaneously with a warning. As an active safety measure that fundamentally prevents the possibility of accidents caused by worker negligence or incorrect judgment, it provides a significantly higher level of accident prevention effectiveness compared to simple warning systems.

[0044] This invention contributes to establishing a sustainable safety culture based on objective data. It evaluates 'system operational reliability' by comparing and analyzing the number of times a system's automatic control is used for manual user intervention. Through this, managers can identify and improve organizational and managerial risks, such as false alarms or worker safety complacency, based on data. This creates a virtuous cycle that leads to the continuous improvement of the safety management system and the establishment of a safety culture in the field.

[0045] The effects according to the present invention are not limited to those exemplified above, and a wider variety of effects are included within the present invention. Brief explanation of the drawing

[0047] Figure 1 illustrates a flowchart between all components according to the present invention. FIG. 2 illustrates a flowchart for determining the risk level of individual workers according to the present invention. Figure 3 illustrates a flowchart of the field unit risk status assessment according to the present invention. Figure 4 illustrates a flowchart for determining a high-altitude work risk situation according to the present invention. FIG. 5 illustrates a flowchart of the system operation reliability evaluation according to the present invention. Specific details for implementing the invention

[0048] Hereinafter, various embodiments are described in more detail with reference to the attached drawings. The embodiments described in this specification may be modified in various ways. Specific embodiments may be depicted in the drawings and described in detail in the detailed description. However, specific embodiments disclosed in the attached drawings are intended only to facilitate understanding of various embodiments. Accordingly, the technical concept is not limited by specific embodiments disclosed in the attached drawings, and it should be understood that it includes all equivalents or substitutions that fall within the spirit and scope of the invention.

[0049] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but these components are not limited by the aforementioned terms. The aforementioned terms are used solely for the purpose of distinguishing one component from another.

[0050] Functions related to artificial intelligence according to the present disclosure are operated through a processor and memory. The processor may be composed of one or more processors. In this case, the one or more processors may be general-purpose processors such as CPUs, APs, and DSPs (Digital Signal Processors), graphics-dedicated processors such as GPUs and VPUs (Vision Processing Units), or artificial intelligence-dedicated processors such as NPUs. The one or more processors control the processing of input data according to predefined operation rules or artificial intelligence models stored in memory. Alternatively, if the one or more processors are artificial intelligence-dedicated processors, the artificial intelligence-dedicated processors may be designed with a hardware structure specialized for processing a specific artificial intelligence model.

[0051] The predefined rules of operation or artificial intelligence models are characterized by being created through learning. Here, being created through learning means that a basic artificial intelligence model is trained using a number of training data by a learning algorithm, thereby creating predefined rules of operation or artificial intelligence models configured to perform desired characteristics (or objectives). Such learning may be performed on the device itself where the artificial intelligence according to the present disclosure is executed, or it may be performed through a separate server and / or system. Examples of learning algorithms include supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but are not limited to the examples described above.

[0052] An artificial intelligence model can be composed of multiple neural network layers. Each of the multiple neural network layers has multiple nodes and weight values, and performs neural network operations through calculations between the results of previous layers and the multiple weights. The multiple weights possessed by the multiple neural network layers can be optimized based on the learning results of the artificial intelligence model. For example, multiple weights can be updated so that the loss value or cost value obtained by the artificial intelligence model during the learning process is reduced or minimized. Additionally, to minimize the loss value or cost value, multiple weights can be updated in a direction that minimizes the gradient associated with the loss value or cost value. Artificial neural networks may include deep neural networks (DNNs), such as Convolutional Neural Networks (CNNs), Deep Neural Networks (DNNs), Recurrent Neural Networks (RNNs), Restricted Boltzmann Machines (RBMs), Deep Belief Networks (DBNs), Bidirectional Recurrent Deep Neural Networks (BRDNNs), or Deep Q-Networks, but are not limited to the examples mentioned above.

[0053] A network is a network that serves as a transmission path for web pages; it may be a closed network such as a LAN (Local Area Network) or WAN (Wide Area Network), but it is desirable for it to be an open network such as the Internet. The Internet refers to a global open computer network structure that provides the TCP / IP protocol and various services existing at its upper layers, namely HTTP (HyperText Transfer Protocol), Telnet, FTP (File Transfer Protocol), DNS (Domain Name System), SMTP (Simple Mail Transfer Protocol), SNMP (Simple Network Management Protocol), NFS (Network File Service), and NIS (Network Information Service).

[0054] Terminals can be implemented in various forms. For example, the terminals described in this specification may include mobile terminals such as smartphones, tablet PCs, PDAs, portable multimedia players, and MP3 players, as well as fixed terminals such as smart TVs and desktop computers.

[0055] In this specification, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof. When a component is described as being “connected” or “connected” to another component, it should be understood that it may be directly connected to or connected to that other component, or that there may be other components in between. On the other hand, when a component is described as being “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between.

[0056] Meanwhile, a "module" or "part" for a component as used in this specification performs at least one function or operation. Furthermore, a "module" or "part" may perform a function or operation by hardware, software, or a combination of hardware and software. Additionally, a plurality of "modules" or a plurality of "parts," excluding a "module" or "part" that must be performed on specific hardware or on at least one processor, may be integrated into at least one module. A singular expression includes a plural expression unless the context clearly indicates otherwise.

[0057] In addition, power, power transmission, and control therefor for the following assembly configurations and embodiments, including "by control," follow conventional technology including terminals, applications, hardware control modules, etc., so they are omitted to avoid redundancy.

[0058] In addition, the operation embodiments and configurations described in a general manner without being explained in detail below follow the prior art and are omitted in order to focus on describing the purpose of the present invention and the resulting effects.

[0059] Furthermore, in describing the present invention, if it is determined that a detailed description of related known functions or configurations may unnecessarily obscure the essence of the invention, such detailed description is abbreviated or omitted.

[0060] The present invention relates to a communication construction management system for preventing safety accidents at a communication construction site. The following describes specific components for implementing the present invention, including a data collection unit (100), a storage unit (200), a control unit (300), and a notification unit (600).

[0061] The above data collection unit (100) refers to a collection of multiple hardware sensor devices distributed at various locations within a telecommunications construction site to measure and collect raw data necessary for safety management in real time, such as the worker's biosignals, environmental conditions, the worker's location and posture, and site video. The above data collection unit (100) acquires data from each detection unit included therein and performs the function of periodically transmitting the acquired raw data or partially pre-processed data to the control unit (300) described later via wired or wireless communication methods, or when an event occurs. In some embodiments, the above data collection unit (100) may include an edge computing device that independently performs basic event filtering or counting functions. The biosignal detection unit, which is a sub-component of the above data collection unit (100), is embedded in a smart safety helmet or safety band worn by the worker and acquires data by measuring the heart rate once per second through a photoplethysmography sensor and measuring the body temperature at one-minute intervals through a contact temperature sensor. The temperature and humidity detection unit is fixedly installed in the main work area of ​​the site and acquires ambient temperature and humidity data at one-minute intervals. The video recording unit acquires video stream data by capturing video of the site or the worker's viewpoint at a speed of 30 frames per second through a fixed CCTV or a camera attached to the worker's helmet. In one embodiment, the data collection unit (100) may be implemented in a form integrated into the worker's safety helmet, in which case the safety helmet is equipped with a biosignal detection unit, a video recording unit, and a sensor for determining the worker's position and altitude, and all collected data is transmitted to the control unit (300) via the site gateway through the built-in communication module.

[0062] The above storage unit (200) refers to a collection of physical storage media or logical databases that permanently or temporarily record and manage all data required for the operation of the system of the present invention. The above storage unit (200) performs the function of storing raw data received from the data collection unit (100), structured data processed by the data analysis module (400), and the final judgment result of the risk judgment module (500) in a time-series form, and also performs the function of storing core information that serves as the judgment criteria of the risk judgment module (500) and providing it upon request. The data to be stored includes raw data logs collected from all detection units, secondary data generated by the data analysis module (400) such as 'number of heart rate surge events', judgment result data such as 'risk' level judgment history, and pre-set risk conditions including a judgment rule table that defines a plurality of input value conditions and judgment results according thereto that are referenced by the risk judgment module (500) to be described later. In one embodiment, the storage unit (200) may be implemented as a combination of a relational database management system built on a cloud server and a large-capacity file storage, wherein structured data is stored in a database table and unstructured data, such as original image data, is stored in the file storage and connected to an index of the database to perform efficient data management.

[0063] The above control unit (300) is a core entity that performs logical processing procedures such as the 'method for determining individual worker risk level' and the 'method for evaluating risk status at the site unit' according to the present invention, and refers to a central processing unit or distributed processing system that oversees the overall data processing, risk determination, and generation of control commands of the system. The above control unit (300) includes a data analysis module (400) and a risk determination module (500) implemented internally in the form of software, receives raw data from the data collection unit (100), transmits it to the data analysis module (400), and transmits the analyzed result back to the risk determination module (500). After the risk determination module (500) determines the final risk status, it performs the function of generating a signal to control the operation of the notification unit (600), which will be described later, according to the result. The data analysis module (400), which is an internal component of the control unit (300), performs the role of extracting and processing quantitative count data that becomes input parameters for the risk judgment module (500) described later, such as the 'number of heart rate surge events', 'total number of stairs climbed', and 'number of frames of excessive leaning', from the received raw data. For example, it receives a continuous stream of raw heart rate data, counts a section satisfying a preset condition as a single 'event', and outputs the number of such events. The risk judgment module (500), which is another internal component, receives multiple count data processed from the data analysis module (400) and determines the final risk level by performing a hierarchical and conditional judgment procedure according to the judgment rule table stored in the storage unit (200). For example, if the 'number of times the seatbelt is unfastened' is 1 or more, it executes a top priority rule that immediately determines it as 'risk' without considering other data.In one embodiment, the control unit (300) may be implemented as a software application running on an industrial PC installed in a field office or on a virtual machine of a remote cloud server, and the application includes an interface for receiving data, a data analysis module (400), a risk determination module (500), and an output interface for sending control signals to an external device.

[0064] The notification unit (600) refers to a collection of multiple hardware output devices that deliver a warning in the form of a visual, auditory, or tactile signals to a worker or manager based on the judgment result of the control unit (300). The notification unit (600) receives a control signal containing a risk level ('caution', 'warning', 'danger') determined by the control unit (300), and performs the function of selectively or in combination operating the lower alarm sound output unit, vibration generation unit, and display unit according to a preset notification matrix. The alarm sound output unit, which is a lower component of the notification unit (600), outputs an alarm sound at 1-second intervals when a 'warning' level signal is received, and a continuous siren when a 'danger' level signal is received. The vibration generation unit generates a short single vibration when a 'caution' level signal is received, and a continuous strong vibration at a 'warning' level or higher. The display unit displays the status in green, yellow, or red on the manager terminal screen according to the risk level, and when a 'danger' level signal is received, it displays the location of the corresponding worker and the cause of the risk together as text. In one embodiment, the notification unit (600) is composed of a combination of a warning light and speaker installed at the site, a worker's smart safety band, a large monitor in the site control room, and a manager's smartphone app. When a 'danger' level is determined, the control unit (300) immediately activates the warning light and speaker, simultaneously sends a strong vibration to the worker's safety band, and performs multiple notifications by sending an emergency alert as a push notification to the manager's smartphone app.

[0065] The control unit (300) of the communication construction management system according to the present invention receives safety-related raw data from the data collection unit (100) and performs a complex safety management process.

[0066] First, to determine the risk level of an individual worker, the present invention performs a hierarchical judgment procedure that prioritizes determining the number of occurrences of fatal risk behaviors, such as the worker unbuckling their seatbelt, and only if this is not applicable, compares the cumulative number of complex risk events, such as abnormal heart rate and failure to wear safety equipment, with a dynamically adjusted threshold.

[0067] Next, to evaluate the overall risk status at the site level, the present invention independently evaluates the physical risk based on the number of uninspected communication poles and open manholes, and the human risk based on the number of high-risk workers and the number of rule violations, and then applies a combination rule to adopt the higher of the two risk levels as the final risk state.

[0068] In addition, the present invention performs a judgment procedure to precisely determine hazardous situations during work at heights by constantly evaluating a basic exposure level representing accumulated fatigue during work, and immediately upgrading the hazardous situation to the highest grade regardless of the previous evaluation when hazardous behavior is detected at the moment it becomes a precursor to a fall, such as excessive leaning or slipping.

[0069] Finally, in order to self-diagnose the effectiveness of the system, the present invention performs a procedure to periodically calculate the ratio of the number of manual user interventions to the number of automatic control executions of the system, and based on the result, evaluate the system operation reliability as 'good', 'poor', or 'unsatisfactory' and report it to the manager.

[0071] The criteria for determining the risk level of an individual worker in this invention go beyond simply comparing continuous biosignals with threshold values; they aim to quantitatively aggregate the 'frequency' and 'severity' of specific risk events occurring at telecommunications construction sites and logically combine them to evaluate the worker's safety status from multiple perspectives. To this end, this invention adopts hierarchical and dynamic judgment criteria. The first criterion, the absolute risk behavior priority judgment criterion, is a standard that prioritizes specific behaviors that can lead to a fatal accident even if they occur only once, thereby making an immediate 'risk' level determination over all other data. The second criterion, the complex risk event accumulation judgment criterion, is a standard for determining situations where the risk level increases when unsafe conditions or behaviors accumulate repeatedly, even if they do not fall under the first criterion. It aggregates the frequency of multiple events, such as biosignal abnormalities, non-compliance with safety equipment, and access to hazardous areas, to subdivide the risk level into categories such as 'caution' and 'warning'. The third criterion, dynamic threshold adjustment, includes a standard that ensures the accuracy and flexibility of judgment by adjusting the judgment threshold used in the second criterion in real time according to field conditions and historical data, rather than using a fixed value.

[0072] The control unit performs a sequential and conditional logical processing procedure to determine the risk level according to the judgment criteria. The technical basis for this lies in industrial safety analysis, which indicates that safety accidents at telecommunications construction sites are caused by complex rather than single factors, and that specific risky behaviors are fatal enough to render all other factors meaningless. Therefore, the present invention enables the most effective safety management with limited resources by adopting a hierarchical and dynamic judgment structure that first filters out the most serious risk factors and then sequentially evaluates accumulated risk factors through situation-adaptive thresholds.

[0073] The data on which the above judgment criteria are based is acquired in real time from each detection unit of the system in the form of the 'number of event occurrences'. The number of heart rate spike events is acquired by counting as one occurrence if the biosignal detection unit monitors the worker's heart rate and it increases by 20 bpm or more compared to the previous 1-minute average within 10 seconds. For example, three occurrences may be acquired during today's work. The number of safety belt disconnections is acquired by counting as one occurrence if a sensor attached to the worker's safety belt lanyard detects that the fastening is disconnected for 2 seconds or more within the high-altitude work area. For example, one occurrence may be acquired during high-altitude work this morning. The number of safety helmet non-wearing episodes is acquired by defining as one 'episode' and counting as one occurrence if the video analysis module remains without a safety helmet for 30 seconds or more. For example, two occurrences may be acquired during material movement this afternoon.

[0074] Results determined as 'Normal', 'Caution', 'Warning', or 'Danger' through the above processing procedure are utilized differently. Results determined as 'Normal' simply record data on the log server without a separate warning. Results determined as 'Caution' trigger a vibration alert on the smart safety equipment worn by the worker. Results determined as 'Warning' send vibration and voice warnings to the worker and simultaneously automatically transmit a warning notification to the on-site safety manager's terminal. Results determined as 'Danger' sound an alarm throughout the entire site, send immediate notifications to workers and managers, and generate an interlock signal to remotely block the operation of aerial work platforms or winches used for the work if necessary.

[0075] The processing procedure of the present invention can be explained through specific scenario examples. Assume a scenario in which a specific worker, who has a history of repeatedly receiving 'caution' warnings during past manhole operations, performs work while the 'Safety Enhancement Mode' at the site is activated due to bad weather. First, the control unit applies dynamic threshold adjustment criteria based on the activation of 'Safety Enhancement Mode' and the worker's past history data, and accordingly, lowers the 'Warning' judgment threshold from the default value of 5 times to 3 times. Second, the control unit collects the number of events from the worker, assuming that the current number of seatbelt unfastenings is 0, the number of heart rate surge events is 2, and the number of episodes of not wearing a safety helmet is 1. Third, the absolute risk behavior priority judgment criteria are applied; since the number of seatbelt unfastenings is 0, it is determined that it is not a 'dangerous' situation, and the process proceeds to the next step. Fourth, the control unit applies the complex risk event accumulation judgment criteria, and the sum of the collected event counts is calculated as 3 times. Fifth, the sum of the above three times is compared with the 'Warning' threshold of three times, which was dynamically lowered in the first step, and since the condition is met, the risk level of the above worker is determined to be 'Warning'. Sixth, the control unit sends vibration and voice warnings to the above worker and automatically sends a detailed notification to the manager terminal stating "Bad weather conditions, worker of history attention, accumulation of complex risks, 'Warning' level reached".

[0076] The logic of the present invention includes measures to ensure reliability and handle exceptional situations. To prevent false positives, the number of times the seatbelt is disconnected includes filtering logic that counts it as a valid event only when a disconnection signal is continuously received for 2 seconds or more to prevent malfunctions caused by momentary communication errors. Additionally, to handle data loss, if there is no data reception from a specific detection unit for 5 minutes or more, the status of the corresponding worker is separately classified as 'unmonitorable' and an inspection notification is sent to the manager so that the absence of data is not mistaken for a 'safe' state.

[0077] The present invention shows a clear difference in operational results compared to the prior art. It assumes the same scenario in which a worker exhibits slight fatigue (two sudden spikes in heart rate) and inattention (one instance of not wearing a safety helmet) in adverse weather conditions. In the case of the prior art using a fixed threshold, the total number of events is three, which falls short of the fixed 'warning' threshold of five, so it is judged only at a 'caution' level, potentially overlooking the severity of the situation. However, the present invention dynamically lowers the 'warning' threshold to three by reflecting situational information regarding adverse weather, and by determining the same three events as a 'warning' level and immediately notifying the manager, it transmits a command to respond preemptively before a potential risk develops into an actual accident.

[0078] The individual worker risk level determination rule table according to the present invention includes the following configuration. The highest priority rule determines the result as 'dangerous' if the number of seatbelt disconnections is one or more. The second priority rule determines the result as 'warning' if the sum of the number of heart rate surge events and the number of helmet non-wearing episodes is above a dynamically adjusted 'warning' judgment threshold and the number of seatbelt disconnections is 0. The third priority rule determines the result as 'caution' if the sum of the above event counts is above a dynamically adjusted 'caution' judgment threshold and below the 'warning' judgment threshold and the number of seatbelt disconnections is 0. The fourth priority rule determines the result as 'normal' if the sum of all event counts is below the 'caution' judgment threshold.

[0079] The simulation example according to the above judgment rule is as follows. It is assumed that the 'Warning' judgment threshold is dynamically adjusted to 3 times. If the number of heart rate surge events is 2, the number of seatbelt unfastening events is 0, and the number of helmet unfastening episodes is 1, the total count becomes 3, and it is judged as 'Warning' according to the second priority rule. If the number of heart rate surge events is 1, the number of seatbelt unfastening events is 1, and the number of helmet unfastening episodes is 0, it is immediately judged as 'Danger' according to the highest priority rule, regardless of the total count.

[0080] The judgment rules presented in this invention, particularly the rule for determining the act of disconnecting a safety belt as the highest priority risk, are technically grounded in the fact that relevant laws and standard safety procedures, such as Article 42 (Prevention of Falls) of the Rules on Occupational Safety and Health, stipulate the wearing of safety harnesses during work at height as the most basic and absolute obligation. Furthermore, since the requirement to strengthen safety standards based on on-site variables such as adverse weather or worker proficiency is a fundamental principle of industrial safety management, dynamically adjusting the judgment threshold in conjunction with this is technically inevitable.

[0081] The components and rule adoption of the present invention provide the following technical effects. By prioritizing absolute hazardous behaviors, it preemptively blocks and prevents risk factors with the highest probability of accidents. By dynamically adjusting judgment criteria based on site conditions and worker history, it identifies potential risks that might be missed with fixed standards, thereby maximizing the accuracy and effectiveness of warnings. Since the risk level of workers is evaluated based on quantitatively counted event counts and clear rule tables rather than the manager's subjective judgment, it enables consistent and objective safety management.

[0082] The terms and threshold values ​​used in the present invention define the criteria as follows. "Work at height" includes work performed in places where there is a risk of falling, such as work platforms, the ends of passageways, or openings, and specifically refers to work performed on communication poles, utility poles, steel towers, etc. The system recognizes a state of work at height when the worker is located at a height of 2 meters or more above the ground through altitude information from a tag worn by the worker. A "heart rate spike event" is counted as one occurrence if the heart rate measured within 10 seconds increases by 20 bpm or more compared to the moving average heart rate over the previous 1 minute. "Dynamic threshold value" is a threshold value used in the judgment rule table above, and includes automatically lowering the threshold value by a certain percentage from the default value when the administrator system sets "Safety Enhancement Mode" or when the system detects through analysis of past data that the frequency of risk warnings for a specific worker exceeds a preset standard.

[0083] In the present invention, the basic threshold value for determining the 'warning' level (e.g., the number of accumulated complex risk events) can be set exemplarily by considering the work type and management response time of the telecommunications construction site. For example, the present invention may set the threshold value by (i) reflecting the repetitive and cumulative characteristics of risk events leading to an actual accident, or (ii) investigating the average response time of site managers (e.g., measuring response times for multiple managers) to align with the time window for alarm propagation and verification. Since these values ​​and criteria may vary depending on the site scale, work process, and telecommunications environment, those skilled in the art may modify and apply the threshold value within the scope of achieving the same objective.

[0084] To set the above 'warning' level judgment threshold (Th_warn), 10 near miss cases recorded in the past database were analyzed, and the cumulative number of complex risk events (rapid increase in heart rate + failure to wear a helmet) during the 1 hour immediately preceding the accident was calculated through simulation.

[0085] [Table 1-1]

[0086]

[0087] As a result of the simulation above, the cumulative number of complex risk events in 10 cases involving near misses was found to be an average of 5.7, a minimum of 4, and a median of 5.5. In particular, it was confirmed that the cumulative number of events was 5 or more in 8 out of 10 cases (80%). Therefore, it can be seen that setting the threshold to 5 is a statistically significant criterion with 80% explanatory power in identifying the interval where the probability of near misses increases rapidly. This proves that the setting of Th_warn = 5 is a reasonable decision based on the distribution of actual risk data.

[0088] To verify the effect of the set threshold (Th_warn = 5), data from 10 intervals (1 hour each) of normal work time at a past site without a system were simulated. In each interval, if the system generated a 'warning', it was considered that a potential risk had been prevented.

[0089] [Table 1-2]

[0090]

[0091] The results of the verification simulation above demonstrate that when the set threshold of 5 is applied, potential risks can be successfully prevented by accurately generating 'warnings' in the four sections (No. 3, 6, 8, and 10) out of a total of 10 work sections that had a high number of accumulated risk events. This proves through simulation that the system's risk prediction accuracy is 100% (4 actual risk sections out of 4 warnings) and the recall rate is 100% (4 warnings out of 4 actual risk sections). If the threshold had been set too high to 9, all risks would have been missed, and if it had been set too low to 2, 8 unnecessary warnings would have occurred. Therefore, it can be seen that the threshold of 5 represents the optimal balance point for effectively identifying actual risk sections while minimizing unnecessary warnings.

[0093] The purpose of the criteria for evaluating the overall risk status at the site level in this invention is to go beyond the sum of individual worker risks and to independently identify inherent physical risk factors and real-time human risk factors occurring at the site, and to organically combine them, thereby evaluating the safety of the entire site from a macroscopic perspective. To this end, this invention adopts dual evaluation and combination criteria. The first criterion, the physical risk assessment criterion, evaluates the static risk level resulting from the management status or physical defects of unique structures and facilities at the telecommunications construction site, such as communication poles and manholes, and identifies potential and environmental risk factors that could cause accidents. The second criterion, the human risk assessment criterion, evaluates the dynamic risk level arising from human activities, such as dangerous behaviors, rule violations, and accumulated fatigue among workers at the site, and assesses the risk level by aggregating unsafe behaviors occurring in real time. The third criterion, the overall risk status determination criterion, includes a standard for comprehensively determining the final risk status of the entire site as 'Safe,' 'Average,' or 'Dangerous' by combining the independently evaluated physical and human risk criterions according to pre-set combination rules.

[0094] The control unit described above performs a logical processing procedure that combines results following parallel evaluation to determine the on-site risk status according to the aforementioned judgment criteria. The technical basis for this lies in the fact that on-site safety is composed of two core axes: equipment safety and worker safety. It is based on the 'weakest point principle,' which holds that if the safety of either side collapses, the safety of the entire system cannot be guaranteed, no matter how safe the other side is. Therefore, the present invention conservatively and reliably evaluates the overall safety status of the site without overlooking any risk, through a dual evaluation and combination structure that independently evaluates the two risk factors and adopts the highest risk level as the final risk state.

[0095] The data on which the above judgment criteria are based is obtained in real-time or periodically from the system's database and each detection unit in the form of 'count' and 'frequency'. The number of uninspected communication poles is obtained by periodically querying the facility management database to determine the total number of communication poles for which one year has passed since the last safety inspection date. For example, two may be obtained at a specific site. The number of open manholes is obtained in real-time by determining the total number of manholes for which an 'open' status signal is received from IoT sensors installed on manhole covers within the site. For example, one may be obtained at a specific site. The number of high-risk workers is obtained by aggregating the total number of workers whose current risk level is determined to be 'Warning' or 'Danger' as a result of the 'Individual Worker Risk Level Judgment Method'. For example, one person may be obtained at a specific site. The total number of rule violation events is obtained by summing the total occurrences of safety rule violation events by all workers recorded in the system over the past hour. For example, three may be obtained at a specific site.

[0096] The overall site risk status determined as 'Safe', 'Normal', or 'Dangerous' through the above processing procedure is utilized differently. A result determined as 'Safe' is displayed in green on the administrator dashboard, and all systems are configured to operate in standard monitoring mode. A result determined as 'Normal' sends a notification to the administrator stating "Site risk level upgraded to 'Normal'" and shortens the data collection and analysis cycle to perform intensive monitoring. A result determined as 'Dangerous' sends an emergency alert stating "Site risk status 'Dangerous' issued!" to all workers' smart safety equipment and the administrator's terminal, along with the key factors that caused the risk, to induce immediate action.

[0097] The processing procedure of the present invention can be explained through a specific scenario example. Assume a specific telecommunications construction site where two telecommunications poles have exceeded their inspection cycle, one manhole is open due to work, all workers are working safely so there are no high-risk workers, and three minor rule violations have occurred. First, the control unit acquires static risk data, and the number of uninspected telecommunications poles is 2, and the number of open manholes is 1. According to the physical risk judgment rule, since the number of open manholes is 1 or more, the physical risk of the site is evaluated as 'high'. Second, the control unit acquires dynamic risk data, and the number of high-risk workers is 0, and the total number of rule violation events is 3. According to the human risk judgment rule, since both conditions satisfy the 'low' criterion, the human risk of the site is evaluated as 'low'. Third, the control unit applies a comprehensive risk status determination rule to combine the evaluated physical risk level ('high') and human risk level ('low'). According to the rule, if either of the two is 'high', the final result is 'dangerous', so the comprehensive risk status of the site is finally determined to be 'dangerous'. Fourth, the control unit sends an emergency alert to the site manager and all workers with specific details, stating, "Site risk status 'Dangerous' issued! Cause: Open manhole (Location: Zone A-3) detected. Exercise caution when approaching the area and take immediate safety measures."

[0098] The logic of the present invention includes measures to ensure reliability and handle exceptional situations. To handle sensor data errors, if communication from a manhole opening / closing sensor is interrupted for more than 10 minutes, the status of the corresponding manhole is not assumed to be 'safe' but is classified as 'status unconfirmed'. If there is one or more facilities with 'status unconfirmed' status, the control unit maintains the overall risk status at least at the 'normal' level and sends a notification requesting a manual inspection to the manager. Additionally, to handle database linkage failures, if the linkage with the facility management database fails and data on uninspected communication poles cannot be updated, the system evaluates the risk level based on the most recently successfully received data, while explicitly warning the manager that more than 24 hours have passed since the data was received to make them aware of the possibility of reduced data reliability.

[0099] The present invention demonstrates a clear difference in operational results compared to the prior art. It assumes the same scenario where all workers at the site fully comply with safety rules and their vital signs are all normal, but a manhole cover located on one side of the work area is open. In the case of the prior art, which performs worker-centric monitoring, it confirms that the condition of all workers is normal and judges the risk level of the site as 'safe' or 'normal,' thereby failing to recognize any fatal physical hazards that workers might fall into while moving. However, the present invention evaluates the human risk level as 'low,' but detects one open manhole via an IoT sensor to evaluate the physical risk level as 'high.' Based on combination rules, it ultimately determines the overall risk status of the site as 'dangerous,' and transmits an alarm by specifically identifying the cause and location of the danger. Consequently, it transmits a command to preemptively block environmental and potential hazards that cannot be detected by human factors alone.

[0100] The physical and human risk assessment rule table according to the present invention includes the following configuration. The physical risk assessment is evaluated as 'high' if the number of open manholes is 1 or more or the number of uninspected communication poles is 3 or more, evaluated as 'medium' if the number of open manholes is 0 and the number of uninspected communication poles is 1 or more but less than 3, and evaluated as 'low' if both the number of open manholes and the number of uninspected communication poles are 0. The human risk assessment is evaluated as 'high' if the number of high-risk workers is 2 or more or the total number of rule violations is 10 or more, evaluated as 'medium' if the number of high-risk workers is 1 or the total number of rule violations is 5 or more but less than 10, and evaluated as 'low' if the number of high-risk workers is 0 and the total number of rule violations is less than 5. In addition, the comprehensive risk status determination rule table combines the aforementioned evaluated physical risk and human risk to determine the final result as 'Dangerous' if either one is rated 'High', 'Safe' only if both are rated 'Low', and 'Average' for all other combinations.

[0101] The judgment rule of the present invention, in particular the rule that evaluates the physical risk level as 'high' when even a single open manhole exists, is based on the fundamental principle of industrial safety that even a single fall accident can be fatal. Likewise, the combination rule that determines the overall risk state as 'dangerous' when either the physical risk level or the human risk level is 'high' ensures objectivity because it is based on the technical necessity that on-site safety can be severely compromised by a problem with only one element.

[0102] The adoption of the components and rules of the present invention provides the following technical effects. By integrally managing not only worker behavior but also physical and environmental factors at the site, it identifies and prevents potential risks that existing systems failed to detect. By clearly distinguishing and presenting the physical and human factors that served as the basis for determining a risk state, it supports safety managers in accurately diagnosing on-site issues and taking customized measures, such as maintenance or safety training. Through combination rules based on the weakest point principle, it ensures that the safety status of the site is always evaluated conservatively, allowing for the management of even minor risk factors without overlooking them.

[0103] The terms and threshold values ​​used in this invention are defined according to the following criteria. "Uninspected communication pole" refers to a communication pole for which 365 days have elapsed since the last structural safety inspection date recorded in the system's facility management database. "High-risk worker" refers to a worker whose real-time risk level is determined to be "Warning" or "Danger" grade through the "Individual Worker Risk Level Determination Method" mentioned above. The presented judgment threshold values ​​are default values ​​set based on accident analysis statistics and standard safety guidelines related to telecommunication construction from the Korea Occupational Safety and Health Agency, and provide a setting function that allows a manager to adjust them according to the scale of the site or the difficulty of the work.

[0104] The setting of the threshold for 'uninspected communication poles' to three in the physical risk assessment criteria of the present invention is based on relevant theories regarding communication infrastructure. While a defect in a single facility is limited to a localized problem, if multiple adjacent facilities are simultaneously in a state of poor management, the possibility that the failure of one facility will lead to a chain of failures increases exponentially. The present invention sets the minimum cluster unit corresponding to the critical mass of such chain failures to three, applying a more conservative standard than the N-2 scenario (a situation where two facilities fail simultaneously) commonly used in the stability analysis of power grids and communication networks.

[0105] To set the threshold value (Th_pole) for the above 'physical risk: high' judgment, 10 hypothetical scenarios regarding the structural stability of communication poles were simulated. In each scenario, the probability of a chain collapse was calculated based on the number of uninspected (assumed potential fault) communication poles.

[0106] [Table 2-1]

[0107]

[0108] This table presents the results of an exemplary simulation for setting physical risk thresholds based on the 'number of uninspected communication poles.' The simulation assumes adjacency relationships and load conditions between communication poles and can calculate the probability of cascading collapse through structural stability analysis (e.g., structural analysis techniques including finite element analysis). However, since the analysis model and input parameters (material properties, boundary conditions, load conditions, etc.) may vary depending on the site facilities and environment, those skilled in the art may apply other structural analysis techniques or safety factor-based evaluations to achieve the same objective.

[0109] Based on the simulation results above, it can be observed that when the number of uninspected communication poles is 1 or 2, the probability of a chain collapse remains very low at 2.1% or less; however, from the moment the number increases to 3 or more, the probability of a chain collapse exceeds 10% and increases rapidly. This implies that 3 uninspected communication poles correspond to a statistical inflection point that seriously threatens the structural stability of the site. Therefore, setting the threshold to 3 is a reasonable and objective criterion for detecting this non-linear pattern of increasing risk.

[0110] To verify the effect of the set threshold (Th_pole = 3), accident data from 10 sites where problems occurred due to aging equipment in the past was simulated.

[0111] [Table 2-2]

[0112]

[0113] The results of the verification simulation above show that, among a total of 10 actual accident site data points, the physical risk judgment rule of the present invention can accurately identify 7 risk situations as 'high' in advance (recall rate 70%). In particular, it successfully identified all 4 cases (No. 1, 4, 8, 10) where there were 3 or more uninspected communication poles and 3 cases (No. 3, 6, 9) where there were open manholes. This demonstrates that the set threshold and rule can effectively detect physical risk factors at the site that led to the actual accident.

[0115] The purpose of the criteria for determining hazardous situations during work at height in this invention is to go beyond the binary fact that a worker is in a high place, to distinguish and evaluate the accumulated fatigue from work at height and instantaneous hazardous behaviors that can lead directly to fall accidents, and to precisely determine the probability of actual fall risk occurrence by hierarchically combining these factors. To this end, this invention adopts hierarchical risk judgment criteria. The first criterion, the basic exposure level evaluation criterion, measures how often and how much a worker ascends and descends high places during working hours to evaluate the basic fatigue and risk exposure levels caused by work at height. The second criterion, the instantaneous hazardous behavior detection criterion, detects in real time behaviors that serve as direct precursors to falls, such as excessive leaning, slipping, and disengagement from scaffolding, which occur during work at height. The third criterion, the risk level escalation judgment criterion, includes a standard for immediately escalating the situation to the highest risk grade if an instantaneous hazardous behavior corresponding to the second criterion is detected even once, regardless of the basic exposure level evaluated according to the first criterion.

[0116] The control unit described above performs a logical processing procedure to determine the hazardous situation of work at height based on the above judgment criteria, constantly evaluating the basic exposure level and immediately raising the alarm level upon the occurrence of a momentary hazardous behavior. The technical basis for this lies in the results of accident analysis indicating that most fall accidents stem from minor carelessness or dangerous postures occurring in a state of accumulated fatigue due to prolonged work at height. Accordingly, the present invention adopts a risk escalation structure that monitors the volume of work and hazardous behavior separately and uses the occurrence of hazardous behavior resulting in the most fatal consequences as the highest priority trigger for the final judgment. This is an effective method that reduces unnecessary alarms to lower alarm fatigue, while enabling a response without delay when an actual risk occurs.

[0117] The data on which the above judgment criteria are based is acquired in real time from the worker's smart safety equipment and video recording unit. The number of times the communication pole climbing starts is acquired by counting it as one time whenever the worker's altitude information changes from less than 2 meters above the ground to 2 meters or more. For example, 4 times may be acquired during morning work. The total number of climbing stairs is acquired by accumulating the count whenever a sensor attached to the worker's foot or ankle detects a unique up-and-down movement pattern of stepping on the communication pole step. For example, a total of 50 stairs may be acquired during morning work. The number of frames of excessive leaning forward is acquired by analyzing video frames to detect the number of frames where the worker's upper body center deviates by more than 50 cm from the communication pole or where a posture of extending the arms by more than 120 degrees relative to the torso is detected. For example, 30 frames may be acquired continuously during a specific task. The number of fall risk motion recognitions is acquired by counting it as one time when the worker's sensor instantaneously detects an impact force of 3G or more or an abnormal change in angular velocity, which is determined to be slipping or stepping off the platform. For example, you can obtain one during a specific task.

[0118] High-altitude work risk situations identified as 'normal exposure', 'high exposure', or 'imminent danger' through the above processing procedure are utilized differently. For results identified as 'normal exposure', only work data is recorded without any separate measures. For results identified as 'high exposure', an audio and vibration alert is sent to the worker stating, "You will be working at heights for a long time. We recommend taking a break shortly," and the manager is notified of the possibility of accumulated fatigue. For results identified as 'imminent danger', a strong audio warning stating, "Danger of falling! Correct your posture immediately!" is simultaneously transmitted to the worker, and an emergency alert is sent to the manager with attached video clips or data logs of the moment the dangerous behavior was detected.

[0119] The processing procedure of the present invention can be explained through a specific scenario example. Assume a situation where a specific worker, who performed work moving between several communication poles in the morning and recorded 4 climbing starts and a total of 50 climbing steps, leans excessively forward to connect a cable during afternoon work. First, the control unit applies the 4 climbing starts and 50 climbing steps accumulated during the morning to the basic exposure level evaluation rule; according to the rule, it initially evaluates the worker's basic exposure level as 'high exposure' and sends a rest recommendation notification. Second, during afternoon work, the video analysis module detects a posture where the worker's upper body center is more than 50 cm away from the communication pole and begins counting the number of frames of excessive leaning. Third, the control unit applies the final danger situation judgment rule; since momentary dangerous behavior has been detected, the criteria for judging an escalation of the danger level are satisfied. Fourth, according to the rule, regardless of the 'high exposure' result evaluated in the first step, the worker's current situation is immediately and finally determined as 'imminent danger'. Fifth, the control unit emits a "Danger of Falling!" voice warning accompanied by a siren sound from the worker's smart safety helmet, and simultaneously sends an emergency notification to the safety manager's terminal, attaching a video clip of the moment the worker leaned out.

[0120] The logic of the present invention includes measures to ensure reliability and handle exceptional situations. To handle video analysis errors, time-based filtering is applied to prevent momentary misrecognition by determining an excessive leaning posture as a valid event only when it persists for 0.5 seconds or longer, in preparation for cases where the recognition of a worker's posture in the video intermittently fails. Additionally, to handle sensor noise, an exception handling logic is included that does not generate a warning if the posture stabilizes within 1 second thereafter, even if an impact force of 3G or more is detected, in order to prevent the worker's everyday movements from being mistaken for actions posing a risk of falling.

[0121] The present invention demonstrates a clear difference in operational results compared to the prior art. It assumes the same scenario where a worker remains safely at a high-altitude work position but leans excessively to the side to grasp a tool. The location-based prior art fails to recognize this as a dangerous situation because the worker's altitude has not changed; it merely displays status information indicating that the worker is "working at a height" and fails to detect the dangerous posture immediately preceding a fall. However, the present invention detects the instantaneous dangerous behavior of "excessive leaning" through video analysis, regardless of the worker's altitude. It immediately raises the risk level to "imminent danger" and transmits an immediate warning to the worker and manager to induce posture correction. Consequently, it accurately captures specific dangerous "behaviors" that could lead to accidents and transmits commands for preventive measures.

[0122] The basic exposure level evaluation rule table for high-altitude work according to the present invention is configured in the form of a matrix with the number of climbing starts and the total number of climbing stairs as two axes, and determines the result as 'normal exposure' or 'high exposure' depending on the combination of the two input values. For example, if the number of climbing starts is between 3 and 5 and the total number of climbing stairs is between 31 and 80, it is determined as 'high exposure'. In addition, the final judgment rule table for high-altitude work risk situations determines the result as 'imminent risk' by making the case where the number of excessive leaning frames is 15 or more or the number of fall risk motion recognitions 1 or more the highest priority condition, and adopts the evaluation result of the basic exposure level evaluation rule table as the final judgment result only when the above conditions are not met.

[0123] The judgment rules of the present invention, particularly the rule that elevates instantaneous dangerous behaviors such as excessive leaning or slipping to 'imminent danger' by prioritizing them over the basic exposure level, are based on the principle of industrial safety engineering that losing body balance during work at height is the most direct and decisive cause of fall accidents. Although accidents do not occur if balance is not lost regardless of the volume of work, a single loss of balance can lead to an immediate accident; therefore, judging these dangerous behaviors as the highest priority is a technical necessity.

[0124] The adoption of the components and rules of the present invention provides the following technical effects. Since the highest level of alarm is sent only when actual dangerous behavior occurs, rather than simply because of a high location, it significantly reduces alarm fatigue caused by workers ignoring alarms and increases the reliability of the alarm. By separately managing the 'high exposure' state and recommending preemptive rest to workers, it performs preventive safety management that reduces the very possibility of accidents caused by accumulated fatigue. When an 'imminent danger' alarm is triggered, specific behavioral data that caused it is recorded along with video, thereby securing objective data that can be used for future accident analysis and safety training materials.

[0125] The terms and threshold values ​​used in this invention define the criteria as follows. "Excessive leaning" refers to a sustained state in which the center point of the worker's upper body, identified through video analysis, deviates horizontally by more than 50 cm from the surface of the supporting communication pole. "Falling risk motion" refers to a case where sudden movement that does not occur within the normal working range is detected by a sensor attached to the worker's body, such as linear acceleration of 3G or more or an angular velocity of 200 deg / s or more. The judgment threshold values ​​presented above are statistical default values ​​set by analyzing the daily average work patterns and risk behavior simulation data of 50 skilled telecommunication construction workers, and can be adjusted by a manager depending on the type of work or the site environment.

[0126] The 50cm deviation of the upper body center, which is the criterion for judging 'excessive leaning' in this invention, is based on ergonomic and biomechanical analysis. Based on the average height of an adult male (175cm), 50cm is the point where the burden applied to the spine and lower limb muscles begins to increase by approximately 1.5 times compared to a vertical posture when the upper body is tilted using the waist as a support point. Since exceeding this point causes a rapid increase in the probability of losing balance even from minor external impacts (e.g., wind, recoil from cables), it is set as an objective biomechanical threshold for judging the risk of falling. Furthermore, the impact force of 3G or more, which is the criterion for judging 'movements at risk of falling,' is a value equivalent to twice the impact force (maximum 1.5G) generated during normal work movements (walking, sitting, etc.), and serves as a clear physical criterion for distinguishing between obvious abnormal situations (slips, falls, etc.) and normal movements.

[0127] To set the above 'excessive body reach' threshold (Th_reach), a high-altitude work posture was simulated using motion capture equipment on 10 subjects, and the failure rate of maintaining balance according to the body reach distance was measured.

[0128] [Table 3-1]

[0129]

[0130] As a result of the simulation above, when the body extension distance was 40 cm or less, the failure rate to maintain balance was stable at less than 5%, but from the moment it reached 50 cm, the failure rate exceeded 10% and showed a tendency to increase rapidly. At 60 cm or more, the failure rate exceeded 25%, resulting in a very dangerous condition. Therefore, it can be seen that setting the threshold to 50 cm is the most reasonable criterion for detecting the inflection point where the risk of losing balance begins to increase substantially.

[0131] To verify the effect of the set threshold (Th_reach = 50cm), 10 videos of near-fall accidents that occurred during past high-altitude work were 3D modeled to analyze the maximum body reach distance of the worker immediately before the accident.

[0132] [Table 3-2]

[0133]

[0134] As a result of the verification simulation above, in the analysis of a total of 10 actual close-range accident videos, the threshold of 50cm of the present invention is shown to accurately identify 7 cases of dangerous leaning behavior as 'imminent danger' before an accident occurs (recall rate 70%). This demonstrates that the set threshold has a very high correlation with the pattern of dangerous behavior that led to actual accidents and proves that it has a substantial effect in preventing fall accidents.

[0136] The criteria for evaluating the operational reliability of the system in this invention aim to go beyond measuring the technical performance of the system itself and to quantitatively assess how effectively the safety system's automated control measures are accepted by workers and managers in the actual field. This is intended to diagnose the effectiveness of the system by analyzing the interaction between humans and the system. To this end, this invention adopts judgment criteria based on comparative analysis. The first criterion, the system automatic control execution criterion, is a standard for tallying the number of times the system detects a risk and executes active control measures, such as actually blocking equipment operation or suspending power, and represents the number of times the system has operated normally as intended. The second criterion, the user manual intervention evaluation criterion, is a standard for tallying the number of times a manager forcibly disables the system's automatic control measures or a worker intentionally ignores a warning sent by the system, and represents human non-compliance or opposing actions against the system's judgment. The third criterion, the comprehensive operational reliability judgment criterion, includes a standard for calculating the relative ratio of the 'number of user manual interventions' to the tallied 'number of automatic control executions' and comparing this ratio with a preset threshold to comprehensively judge the system's current operational reliability as 'good,' 'degraded,' or 'poor.'

[0137] The control unit described above performs a logical processing procedure that periodically collects and compares two opposing behavioral data to evaluate the operational reliability of the system according to the aforementioned judgment criteria. The technical basis for this lies in the fact that the ultimate success of a safety system is determined by the degree of compliance in human-machine interaction; no matter how technically perfect a system may be, it loses its meaning if the user does not trust it, ignores it, or disables it. Therefore, the present invention adopts a structure that compares and analyzes the system's compliant operation and non-compliant response, which corresponds to an essential 'inspection' step for early detection of potential system problems or user complacency regarding safety and for taking corrective measures.

[0138] The data on which the above judgment criteria are based is periodically obtained from the system's control log and notification history database. The number of times the system's automatic control is executed is obtained by aggregating the total number of times the control unit actually generated a signal to cut off equipment operation or power based on results such as a 'danger' level determination. For example, 5 times may be obtained over the past 24 hours. The number of manual releases is obtained by extracting from the system log the total number of times an administrator-level user forcibly released the control measure through an authentication process while the automatic control measure was executed. For example, 1 time may be obtained over the past 24 hours. The number of ignored alarms is obtained by aggregating the total number of alarms for which the operator did not send a response signal, such as pressing an 'OK' button, within the set maximum response time after an alarm of a 'warning' or 'danger' level was issued to the operator. For example, 2 times may be obtained over the past 24 hours.

[0139] System operational reliability, determined as 'Good', 'Degraded', or 'Poor' through the above processing procedure, is utilized differently. A result determined as 'Good' signifies that the system is operating stably, and this status is recorded in the monthly safety report. A result determined as 'Degraded' sends a "System Operational Reliability 'Degraded' Detected" alert to the General Safety Manager, and provides a list of workers who frequently manually deactivated or ignored alarms, along with related logs, to facilitate root cause analysis. A result determined as 'Poor' sends an urgent alert to the General Safety Manager, and if the ignore rate for a specific system alarm is abnormally high, automatically generates and sends a message recommending a review and correction of the sensitivity parameters for that alarm.

[0140] The processing procedure of the present invention can be explained through a specific scenario example. Assume a situation in which the reliability of system operation over the past 24 hours is evaluated at a specific telecommunications construction site. First, the control unit analyzes the system logs to confirm that the operation of the aerial work platform was automatically cut off 3 times and the power to the live section was cut off 2 times, and counts the total number of automatic system control executions as 5. Second, the control unit analyzes the logs for the same period to confirm that the manager manually released the automatically cut-off aerial work platform for emergency movement 1 time and that workers did not respond to warning alarms 2 times, and counts the total number of manual user interventions as 3. Third, the control unit applies the operational reliability judgment rule and calculates the 'manual user intervention rate' as 60%. Since the calculated intervention rate of 60% falls within the 'degradation' range, the system operation reliability of the site is finally determined to be 'degraded'. Fourth, the control unit sends a notification to the overall safety manager stating, "Site D, system reliability judged as 'degraded'. (3 manual interventions compared to 5 automatic control attempts, intervention rate 60%). Check detailed logs and analyze the cause."

[0141] The logic of the present invention includes measures to ensure reliability and handle exceptional situations. When an administrator attempts manual disabling, the system compels the administrator to select a predefined reason or input it directly. Furthermore, during reliability evaluation, disabling cases recorded with legitimate reasons are excluded from analysis or assigned a lower weight to enhance the accuracy of the evaluation. Additionally, the system includes exception handling logic that, if the same type of alarm is repeatedly ignored within a short period by the same worker or in the same area, classifies this not merely as an increase in the number of interventions but as a separate 'alarm sensitivity anomaly pattern,' thereby suggesting to the administrator that the relevant sensor or rule be inspected.

[0142] The present invention demonstrates a clear difference in operational results compared to the prior art. It assumes the same scenario where the system executed automatic control 10 times over a week, and the manager manually deactivated it 8 of those times on the grounds that it "interfered with work." The prior art, which uses an event log recording method, simply records individual events such as "10 automatic control occurrences" and "8 manual deactivation occurrences" in chronological order; unless the manager directly analyzes the logs, it is impossible to recognize the serious condition that the system is effectively neutralized. However, the present invention automatically compares and analyzes the two data sets to calculate that the "user manual intervention rate" is 80%, determines the system operation reliability as "poor" according to judgment rules, and transmits an alert containing the analyzed information to the overall safety manager stating, "System reliability 'poor', 80% of automatic control neutralized. Immediate action required." Consequently, it processes simple data into meaningful information to transmit a command that preemptively reveals potential management issues.

[0143] The system operation reliability judgment rule table according to the present invention includes the following configuration. If the user manual intervention rate is 30% or less, it is judged as 'good'; if it exceeds 30% and is 70% or less, it is judged as 'deteriorated'; and if it exceeds 70%, it is judged as 'poor'. At this time, the user manual intervention rate is calculated by dividing the total number of user manual interventions by the total number of system automatic control executions, and if the number of automatic control executions is 0, it is processed as 'evaluation pending'.

[0144] The reason the judgment rule of the present invention, in particular, uses the ratio of manual intervention to automatic control frequency as a measure of reliability is that it is an objective indicator for measuring the extent to which the system's intended operation is rejected by the user. Low system reliability implies a fundamental problem in system design or safety management, and quantitatively evaluating and managing this constitutes a technical necessity for establishing a sustainable safety management system.

[0145] The adoption of the components and rules of the present invention provides the following technical effects. It proactively identifies organizational and managerial risks, such as attempts to disable the system or complacency regarding safety, which are difficult to detect through individual incident logs alone. By providing data on specific types of warnings that are repeatedly ignored or dismissed, it establishes objective grounds for determining whether alarm thresholds are set too sensitively and for improving the system. Through a process of periodically evaluating system reliability and addressing issues, it establishes a virtuous cycle of safety management that enhances system accuracy, earns user trust, and consequently reduces the rate of manual intervention.

[0146] The terms and thresholds used in this invention are defined according to the following criteria. "Manual release" refers to the act of a user with administrator-level authority or higher logging into the system to forcibly restore the operation or power cutoff status of equipment that is running automatically to a normal state. "Ignored alarm" refers to a case where a worker does not input an "acknowledgement" or "recognition" signal on their terminal within the maximum response time set by the system regarding an alarm of a "warning" or "danger" level transmitted to the worker. The threshold for the user manual intervention rate presented above is a default value set based on the point where a decrease in reliability begins to show a significant correlation with an increase in the actual near-accident rate, determined by statistically analyzing data logs collected during a beta test period targeting multiple telecommunications construction sites, and can be adjusted by the manager according to the safety management goal level of the project.

[0147] In the present invention, the operational reliability judgment threshold (e.g., Th_degrade) can be set as a criterion for distinguishing sections requiring operational inspection based on the ratio between controls automatically performed by the system and manual intervention by a user (administrator). For example, if the manual intervention rate increases to a certain level (e.g., 30% or more), it may be judged as 'degraded' as there is a possibility that the consistency of system application has deteriorated; if it increases to a higher level (e.g., 70% or more), it may be judged as 'poor' as there is a possibility that problems with system operation policies or user safety compliance status have worsened. The above figures are examples and may be changed according to field policies and safety target levels.

[0148] To set the above reliability 'degrade' threshold (Th_degrade), the relationship between the manual user intervention rate and the number of monthly near-miss incidents was simulated for 10 virtual sites.

[0149] [Table 4-1]

[0150]

[0151] The simulation results above showed a clear correlation: when the manual user intervention rate was less than 30%, the number of monthly near-miss incidents remained low at 3 or fewer; however, from the moment the intervention rate exceeded 30%, the number of near-miss incidents increased sharply to 5 or more. This suggests that a manual intervention rate of 30% is a statistical threshold at which the actual safety level at the site begins to deteriorate. Therefore, setting the threshold for judging 'deterioration' at 30% is a reasonable decision based on data.

[0152] To verify the effect of the set threshold (Th_degrade = 30%), the end-of-month manual intervention rate of 10 sites with historical operational data was compared with the actual occurrence of accidents in the following month.

[0153] [Table 4-2]

[0154]

[0155] As a result of the verification simulation above, the manual intervention rate at the end of the previous month exceeded 30% for all four sites (L, N, P, R) where actual accidents occurred in the following month, and the threshold value of the present invention demonstrates that it can predict risk signs at these four sites with 100% accuracy (recall rate 100%). On the other hand, the intervention rate at all six sites where no accidents occurred was less than 30% (specificity 100%). This proves that the set threshold value of 30% functions as a highly reliable leading indicator for predicting the likelihood of future accidents.

[0157] The entire operation process of a telecommunications construction management system according to one embodiment of the present invention can be explained through a specific scenario, and this embodiment shows how the data collection unit, storage unit, control unit, and notification unit defined above, and processes 1 to 4 performed by the control unit interact to manage safety at the site in a three-dimensional manner.

[0158] When a manager starts work at the site and activates the system on the morning of the workday, the storage unit contains pre-stored information on the site's equipment and workers. At this time, it is assumed that the equipment information includes two uninspected communication poles, and the history information of a specific worker contains a record of having received 'caution' warnings three or more times in the past. Upon activation, the control unit begins to evaluate the initial risk state of the site and performs a site-unit risk state evaluation method. The risk judgment module of the control unit obtains data from the storage unit indicating that there are two uninspected communication poles, evaluates the physical risk level of the site as 'medium' according to the judgment rule table, and determines the overall risk state of the site as 'normal' according to the combination rule table because the current human risk level is 'low'. Simultaneously, the risk judgment module applies the dynamic threshold adjustment criteria of the individual worker risk level judgment method based on the worker's past warning history, and accordingly, lowers the 'warning' judgment threshold for complex risk events for the worker from the default value of 5 times to 3 times, and records this adjusted threshold in the temporary memory area of ​​the storage unit. On the administrator dashboard, which is the display part of the above notification unit, the status of the site is displayed in yellow ('normal'), and information such as "physical risk level 'medium' (cause: 2 uninspected communication poles)" is also displayed.

[0159] When a specific worker begins cable connection work on a communication pole using an aerial work platform during the morning, a data collection unit attached to the worker's safety helmet detects the worker's ascent via an altitude sensor, and a sensor on the ankle detects the movement of stepping on a platform to climb. This raw data is transmitted to the control unit, and the data analysis module of the control unit processes this data to tally the 'number of starts climbing the communication pole' as 1 and the 'total number of steps climbed' as 15. Subsequently, the risk judgment module receives the analyzed data and performs a method to determine the risk situation of aerial work, and based on a judgment rule table, initially evaluates the worker's basic exposure level as 'normal exposure'. At this stage, no separate warning is generated, and the corresponding work log is recorded in the storage unit.

[0160] Subsequently, when the worker leans excessively outward while leaning against the railing of the aerial work platform basket because the necessary tool is out of reach, the video recording unit, which is the worker's helmet camera, captures this scene and transmits the video stream to the control unit in real time. The data analysis module detects, through a video analysis algorithm, that the worker's upper body center has deviated more than 50 cm from the support and counts the 'number of frames of excessive leaning out' to confirm that it has lasted for 1 second. Since 'excessive leaning out' has been detected, the risk judgment module applies the risk level escalation judgment criteria of the aerial work risk situation judgment method and immediately determines the worker's aerial work risk situation as 'imminent risk' regardless of the initial evaluation result of 'normal exposure'. At the same time, this 'excessive leaning out' event is also considered as a personal risk event for the worker, and one is added to the cumulative count of his complex risk events.

[0161] As soon as the above-mentioned risk judgment module determines 'imminent danger,' the above-mentioned control unit simultaneously generates control and notification commands. First, a control signal stating "temporarily block ascent and movement functions" is transmitted to the control system of the aerial work vehicle to automatically stop the movement of the aerial work vehicle, and this record of automatic control execution is stored as a log in the above-mentioned storage unit, and the 'number of system automatic control executions' is increased by one. Next, a multi-notification command is transmitted to the above-mentioned notification unit to cause a strong voice warning, "Danger of falling! Immediately correct your posture!" accompanied by a siren sound, to be emitted from the speaker built into the worker's safety helmet, and at the same time, an emergency push notification is sent to the manager's smartphone with an attached video clip of the moment the body was leaned out.

[0162] After assessing the situation, the site manager instructs the worker to correct their posture and then manually disables the system's automatic blocking function in the manager app to move the work vehicle slightly. The manager's manual disable command is authenticated and processed by the control unit, and this action is counted as a 'manual disable count' and recorded in the storage unit, along with the reason for disablement, 'urgent movement required'. This increases the 'user manual intervention count' by one.

[0163] At the end of the work or at a period designated by the manager, the control unit performs a comprehensive evaluation based on the accumulated data. In the final evaluation of the individual worker risk level determination method, the worker has one 'imminent risk' event and one other 'caution' event, resulting in a total of two accumulated complex risk events. Although this does not reach the dynamically adjusted 'warning' threshold (3 times), the system classifies the worker as 'today's high-risk worker' because of the history of 'imminent risk'. Subsequently, in the final evaluation of the site-unit risk status evaluation method, since one 'high-risk worker' has been identified, the human risk level of the site is adjusted upward to 'medium'. The physical risk level of 'medium' and the human risk level of 'medium' are combined, so the overall risk status of the site remains 'normal', but it is recorded that the cause of the risk has changed. Finally, in the final evaluation of the system operation reliability evaluation method, the total number of 'system automatic control executions' over the past 24 hours is tallied as 1, the number of 'user manual interventions' is tallied as 1, and the 'user manual intervention rate' is 100%, so the system operation reliability is judged as 'poor' according to the judgment rule table.

[0164] Finally, a "Daily Safety Report" is generated on the administrator dashboard, which is the display part of the aforementioned notification unit, providing analyzed information such as "High-risk worker: Worker A (Reason: Excessive leaning during work at height)," "Site risk status: Normal (Reason: Increased human risk)," and "System reliability: Poor (Reason: 100% manual disabling of automatic control)," thereby presenting objective grounds for establishing the safety plan for the following day. As described above, one embodiment of the present invention demonstrates that each component and a multi-layered judgment process are organically linked to implement a comprehensive safety management cycle that goes beyond simple detection to encompass prediction, real-time response, and post-evaluation.

[0165] Although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. Various modifications are possible by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention. Explanation of the symbols

[0166] Data collection unit (100) Storage unit (200) Control unit (300) Data analysis module (400) Risk judgment module (500) Notification unit (600)

Claims

Claim 1 A telecommunications construction management system comprises: a data collection unit that collects raw safety-related data from workers and the site; a storage unit that stores preset risk conditions including a judgment rule table; a control unit that receives the raw data from the data collection unit, determines a risk level based on the judgment rule table stored in the storage unit, and generates a control signal according to the judgment result; and a notification unit that transmits a warning to a worker or manager according to the control signal of the control unit, wherein the control unit includes a data analysis module that processes the raw data received from the biosignal detection unit, the position / posture motion detection unit, and the image capturing unit of the data collection unit to generate quantitative count data of risk events, including the number of heart rate surge events in which the heart rate increases above a preset threshold value within a preset time, the number of times a safety belt is detached within a high-altitude work zone, and the number of episodes of not wearing a safety helmet. A telecommunications construction management system comprising a risk judgment module that receives quantitative count data from the data analysis module and determines a final risk level by applying the judgment rule table, wherein the risk judgment module separately classifies the status of the worker as 'unable to monitor' and sends an inspection notification to the manager when there is no data reception from a specific detection unit for more than 5 minutes, thereby processing to prevent the absence of data from being mistaken for a 'safe' state, and applies a dynamic threshold adjustment standard that automatically lowers the judgment threshold compared with the quantitative count data in the judgment rule table from a preset default value based on the activation of the site's 'safety reinforcement mode' due to bad weather or the history of the past warning occurrence frequency of a specific worker. Claim 2 A telecommunications construction management system according to claim 1, wherein the risk judgment module performs a hierarchical judgment structure comprising: a first rule for determining the risk level of an individual worker as 'risk' immediately regardless of the value of other count data when the number of times the seat belt is disconnected is one or more; and a second rule for determining the risk level of the worker as 'warning' or 'caution' when the first rule does not apply, by comparing the sum of the number of heart rate surge events and the number of episodes of not wearing a safety helmet with the first threshold value stored in the storage unit and automatically lowered. Claim 3 A telecommunications construction management system according to claim 1, wherein the risk judgment module, in order to determine the risk situation of the high-altitude work, acquires the number of telecommunications pole climbing starts and the total number of climbing stairs from the data analysis module and evaluates the basic exposure level of the high-altitude work; a standard for detecting in real time the number of frames of excessive leaning or the number of fall risk motions recognized from the video capturing unit or sensor; and a risk level escalation judgment standard that applies the risk level escalation judgment standard to immediately escalate the situation to the highest risk grade, 'imminent risk,' if even a single instance of instantaneous dangerous behavior corresponding to the detected standard is detected, regardless of the evaluated basic exposure level. Claim 4 A communication construction management system according to paragraph 2, wherein the risk judgment module evaluates the physical risk level of the site by obtaining the number of uninspected communication poles for which a preset period has elapsed since the last safety inspection date from the facility management database stored in the storage unit to evaluate the overall risk status at the site level, and by obtaining the number of manholes with open lids from IoT sensors installed at the site, wherein if communication from the manhole opening / closing sensor is interrupted for more than 10 minutes, the status of the corresponding manhole is classified as 'status unconfirmed', and if there is one or more facilities with 'status unconfirmed', the overall risk status is maintained at least at the 'normal' grade level or higher, and the human risk level of the site is evaluated by aggregating the number of high-risk workers judged as 'warning' or 'danger' according to the judgment result and the total number of rule violation events recorded in the system. Claim 5 A telecommunications construction management system according to claim 4, wherein the risk judgment module independently evaluates the physical risk level and the human risk level as 'high', 'medium', and 'low' according to judgment rules stored in the storage unit, respectively; and applies a combination rule to determine the overall risk status of the site as 'risk' regardless of the evaluation result of the other when either the evaluated physical risk level or the human risk level is evaluated as 'high'. Claim 6 In paragraph 2, the control unit, when the risk judgment module determines the risk level of the worker as 'risk' according to the first rule, controls the notification unit to perform multiple notifications by simultaneously transmitting voice and vibration warnings to the worker and sending an emergency notification to the manager, and further prevents a fall accident by generating and transmitting an interlock signal to remotely block the operation of the aerial work vehicle or winch being used by the worker in whom the risk is detected, and the control unit, in order to self-diagnose the effectiveness of the system, aggregates the number of times the system automatically blocks the operation of the equipment including the generation of the interlock signal and the number of times the manager forcibly disables the automatic control measure or the worker ignores the warning alarm, respectively, calculates the user manual intervention rate, which is the ratio of the number of times the user manual intervention is performed to the aggregated number of times the system automatically controls the operation, and comprehensively determines the current operational reliability of the system as 'good', 'deteriorated', or 'poor' by comparing the calculated rate with a preset threshold value.

Images