COMPREHENSIVE AIoT BASED SAFETY MONITORING SYSTEM FOR INDUSTRIAL WORKERS
The AIoT-based system integrates sensors and devices to analyze IoT data for real-time risk detection and tracking, addressing the limitations of independent systems by enhancing safety monitoring and reducing false alarms.
Patent Information
- Application Number
- KR1020250203106
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2025-12-18
- Publication Date
- 2026-11-06
- Estimated Expiration
- 2045-12-18
AI Technical Summary
Existing industrial site worker safety monitoring systems operate independently, lacking data integration and fused analysis, which hinders comprehensive real-time judgments and automated responses to complex risk situations, leading to potential safety accidents.
A complex AIoT-based system that integrates various sensors and devices into a single embedded board, using AI algorithms to analyze IoT data from environmental, video, and audio sources, and autonomously track worker movements, detecting anomalies and risks in real-time while protecting personal information.
The system effectively detects complex risks such as worker collapse, high fever, and hazardous environments, reducing false positives and negatives, enabling proactive safety measures and predictive maintenance, thus enhancing worker safety and environmental monitoring.
Smart Images

Figure 112025143515684-PAT00001_ABST
Abstract
Description
Technology Field
[0001] The present invention relates to a complex AIoT-based industrial site worker safety monitoring system. By analyzing various IoT sensor information using artificial intelligence to determine the occurrence and likelihood of dangerous situations and environments, the system enables rapid on-site response and prompt dissemination of information, and provides a complex AIoT-based industrial site worker safety monitoring system capable of preemptively preventing safety accidents. Background Technology
[0003] Various monitoring systems and technologies are being developed and utilized in industrial sites to ensure worker safety. These technologies can be broadly classified into sensor-based detection systems, video surveillance systems, location tracking systems, and wearable safety devices.
[0004] Sensor-based detection systems are the most commonly used safety monitoring technology in industrial settings. Individual sensors, such as gas detection sensors, temperature sensors, vibration sensors, and pressure sensors, detect specific risk factors and transmit warning signals to managers. These sensors are typically installed at fixed locations and operate by monitoring environmental changes at those points.
[0005] Furthermore, video surveillance systems are technologies that utilize CCTV cameras to monitor industrial sites in real time. Managers can check the situation on-site via multiple monitors from a central control room and review recorded footage afterwards. Recently, the widespread adoption of high-resolution IP cameras has made it possible to obtain clearer video, and it has become possible to view footage in real time from remote locations via smartphones or PCs. Of course, some systems also incorporate AI technology to provide features that automatically detect situations such as failure to wear safety helmets or entering hazardous areas.
[0006] A location tracking system is a system that identifies a worker's location in real time by utilizing technologies such as RFID tags, beacons, and GPS. It monitors whether workers enter hazardous areas through tags or badges carried by the workers and can quickly determine their exact location in the event of an emergency. It is particularly useful in environments where multiple workers operate simultaneously in large spaces, such as construction sites or large-scale manufacturing facilities.
[0007] Wearable safety devices are safety monitoring equipment worn by workers, and products integrating sensors and communication functions into safety helmets, safety belts, vests, and the like are being released. These devices monitor workers' biometric information, such as heart rate, body temperature, and fall detection, as well as behavioral patterns in real time, sending immediate warnings in the event of abnormal signs. Some smart helmets are equipped with various safety features, including wear detection, fall detection, and panic buttons.
[0008] In addition to this, integrated safety management platforms based on IoT technology are emerging. These platforms collect various types of sensor data into a single system, manage data via the cloud, and provide an environment that allows for comprehensive monitoring of on-site conditions through dashboards.
[0009] In the case of video detection systems, it is currently difficult to acquire video information of workers who are out of the camera's field of view because fixed CCTVs are used. Furthermore, due to simple video recording, it is difficult to accurately determine whether workers are properly wearing the safety equipment that is essential for their safety, and protecting personal information in the event of video leakage is also challenging.
[0010] Furthermore, existing safety monitoring systems have structural limitations as individual sensors or devices operate independently and are optimized for a single function. Sensor-based detection technologies measure only specific items such as gas, temperature, and vibration; video surveillance systems provide only visual information; and location tracking systems identify only the location of workers; thus, each system is separated. Due to this independent configuration, data integration and fused analysis between systems cannot be achieved, making it difficult to make comprehensive real-time judgments and automated responses to complex risk situations occurring across various process lines in industrial sites. Even if subtle anomalies occur in individual systems, they are not linked with information from other systems, resulting in limitations that prevent early detection and response. Consequently, there are fundamental constraints on the proactive prevention of major accidents. Prior art literature
[0012] (Patent Document 1) Korean Registered Patent Publication No. 10-2519988 (Patent Document 2) Korean Registered Patent Publication No. 10-2088457 The problem to be solved
[0013] The present invention has been devised to solve the aforementioned problems and provides a complex AIoT-based industrial site worker safety monitoring system that integrates various sensors and devices into a single embedded board. Through this technology, it automatically detects abnormal and dangerous situations such as worker collapse, high fever, helmet and mask wearing status, fire, dangerous sounds, and exposure to hazardous environments by analyzing various IoT sensors and video and audio data in real time using AI algorithms. It also automatically tracks worker movement through autonomous thermal imaging camera processing technology, enables the analysis of high fever conditions and worker behavior while minimizing the exposure of personal information, thereby simultaneously satisfying personal information protection and safe monitoring. Furthermore, it can compensate for errors caused by false positives or non-detections of some sensors, prevent serious accidents that may occur in industrial sites in advance, and guarantee worker safety in real time, thereby improving the overall safety of the work environment. means of solving the problem
[0015] The present invention provides a complex AIoT-based industrial site worker safety monitoring system comprising a field terminal that senses and detects on-site safety using various IoT sensors and equipment according to the present invention and analyzes the occurrence of risks based thereon, a server unit that monitors the analysis results of the field terminal, and a mobile terminal unit that receives notification results from the server unit.
[0016] The above-mentioned field terminal includes a terminal sensing unit that senses the industrial site, a terminal integrated analysis unit that analyzes whether a risk has occurred based on the sensed information, a terminal communication unit that communicates with a server unit to provide analysis results and receives external control signals, and a terminal control unit that controls the operation of each unit according to the control signals.
[0017] The above terminal sensing unit includes an environmental sensor unit that senses environmental information of the site, a smart video recording unit that records the site conditions and the actions of the worker, a thermal imaging unit that records the heat distribution of the site and the worker, and an acoustic sensor unit that detects abnormal sounds and danger signals.
[0018] The above-mentioned environmental sensor unit uses temperature, humidity, fine dust, and TVOC multi-sensors, as well as CO2 sensors, carbon monoxide sensors, LPG and LNG gas sensors, and formaldehyde sensors; the above-mentioned smart video capturing unit includes a camera unit that captures video, an actuation unit that moves the camera unit according to an image processing module, an image processing module that detects a worker within the captured video and determines the movement of the worker to track the worker through the camera unit, and an image storage module that stores the video captured by the camera unit; and the above-mentioned thermal imaging unit includes a thermal imaging camera unit that captures thermal images, a thermal imaging actuation unit that moves the thermal imaging camera unit, a thermal imaging processing module that detects a worker within the thermal imaging information and determines the movement of the detected worker to control the movement of the thermal imaging camera unit, and a thermal imaging storage unit that stores thermal imaging information captured by the thermal imaging camera unit.
[0019] The above-mentioned terminal integrated analysis unit includes an environment analysis module that analyzes the environmental conditions of the site, an image analysis module that analyzes captured images, a thermal image analysis module that analyzes thermal images, an acoustic analysis module that analyzes acoustics, and an integrated judgment unit that determines whether abnormalities or risks occur based on the analysis results of each analysis module.
[0020] The aforementioned integrated judgment unit simultaneously performs data fusion and situational awareness, calculates a final composite risk index by synthesizing the results of surrounding environment analysis, image analysis, thermal imaging analysis, and acoustic analysis, and determines the final risk situation and response stage by considering reliability and the timing of occurrence.
[0021] In addition, the present invention provides a complex AIoT-based industrial site worker safety monitoring system comprising: an equipment detection unit for detecting abnormalities in equipment installed at an industrial site according to the present invention; a site detection unit for detecting workers and the environment of the industrial site; an integrated judgment unit for determining whether a risk has occurred using the detection results of the equipment detection unit and the site detection unit; a server unit for monitoring information of the integrated judgment unit; and a site control unit connected to the integrated judgment unit and managing the environment of the industrial site.
[0022] The integrated judgment unit includes an equipment analysis unit that analyzes equipment abnormalities based on the detection results of the equipment detection unit, a field analysis unit that analyzes worker and environmental abnormalities based on the detection results of the field detection unit, an integrated analysis unit that analyzes and judges dangerous situations based on the analysis results of the equipment analysis unit and the field analysis unit, an integrated control unit that controls the operation of each unit, and an integrated communication unit that communicates with a server.
[0023] The above-mentioned equipment analysis unit includes a vibration analysis module for analyzing vibration abnormalities of the equipment, an electrical analysis module for analyzing electrical characteristic abnormalities of the equipment, a temperature analysis module for analyzing temperature abnormalities of the equipment, an acoustic analysis module for analyzing acoustic abnormalities of the equipment, and a degradation analysis module for analyzing whether the equipment is degraded, and the above-mentioned field analysis unit includes an environment analysis module for analyzing environmental conditions of the field, an image analysis module for analyzing captured images, a thermal image analysis module for analyzing thermal images, and an acoustic analysis module for analyzing acoustics. Effects of the invention
[0025] As such, the present invention integrates and analyzes environmental, video, thermal, acoustic, and equipment status data in real time on a single embedded-based AIoT platform, thereby automatically detecting complex risk situations at an early stage, such as worker collapse, high heat, failure to wear safety equipment, fire, and exposure to hazardous environments, and preventing risks in advance.
[0026] In addition, the present invention enables worker tracking and behavior analysis that is robust against disturbances such as lighting changes, occlusion, and environmental noise through a structure that separates and analyzes general images and thermal images and then makes an integrated judgment, and can effectively reduce false positives and false negatives compared to a single-sensor-based system.
[0027] In addition, by applying autonomous thermal imaging processing technology and image analysis based on personal information anonymization, it is possible to simultaneously satisfy safety monitoring and personal information protection by minimizing the exposure of personal information while tracking worker movements and analyzing high-temperature conditions and hazardous behaviors.
[0028] In addition, by utilizing equipment anomaly detection and deterioration tracking functions based on vibration, current, temperature, sound, etc., signs of equipment failure can be recognized early and predictive maintenance can be enabled, thereby preventing secondary safety accidents that may occur from equipment anomalies in advance. Brief explanation of the drawing
[0030] FIG. 1 is a conceptual diagram illustrating a composite AIoT-based industrial site worker safety monitoring system according to one embodiment of the present invention. FIG. 2 is a conceptual diagram for explaining a field terminal unit according to one embodiment. FIG. 3 is a block diagram illustrating a terminal sensing unit according to one embodiment. FIG. 4 is a block concept diagram for explaining a smart video capturing unit according to one embodiment. FIG. 5 is a block diagram illustrating a thermal imaging unit according to one embodiment. FIG. 6 is a block diagram illustrating a terminal integration analysis unit according to one embodiment. FIG. 7 is a conceptual diagram illustrating a composite AIoT-based industrial site worker safety monitoring system according to one variant of the present invention. FIG. 8 is a block diagram illustrating an integrated judgment unit according to one variant example. FIG. 9 is a block diagram illustrating an equipment analysis section according to one modified example. FIG. 10 is a block diagram illustrating a field analysis section according to one variant example. Specific details for implementing the invention
[0031] Hereinafter, embodiments of the present invention will be described in more detail with reference to the attached drawings. However, the present invention is not limited to the embodiments disclosed below but may be implemented in various different forms; these embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention. Identical reference numerals in the drawings refer to identical elements.
[0032] It is intended to clarify that the classification of components in this specification is merely based on the primary function each component is responsible for. That is, two or more components described below may be combined into a single component, or a single component may be divided into two or more components based on more subdivided functions. Furthermore, each component described below may additionally perform some or all of the functions performed by other components in addition to its own primary function, and it is obvious that some of the primary functions performed by each component may be exclusively handled by other components. Therefore, the existence of each component described in this specification should be interpreted functionally. For this reason, it is clearly stated that the configuration of the components of the composite AIoT-based industrial site worker safety monitoring system of the present invention may vary to the extent that the objectives of the present invention can be achieved.
[0033] In this specification, relational terms such as first and second, upper and lower, etc., may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying an actual relationship or order between such entities or actions. The terms “comprises,” “comprising,” or other variations thereof are intended to cover non-exclusive inclusions so that a process, method, product, or device comprising a list of components may not include only the component but may include other components not explicitly enumerated or inherent in such process, method, product, or device. A component proceeding to “comprising” excludes, without further limitation, the presence of additional identical components within the process, method, product, or device comprising the component.
[0034] The specific features of the present invention and the method for implementing them are described in detail with respect to an online marketing automation system and method together with the attached drawings.
[0035] However, the drawings and detailed description have briefly described or omitted configurations and methods that are ordinarily known to those skilled in the field of complex AIoT-based industrial site worker safety monitoring systems, and have been limited to the parts necessary for a clear understanding of the disclosure of the present invention. Accordingly, the present invention is not limited to the embodiments disclosed below and can be implemented in various forms and methods, and the present invention is defined by the scope of the claims.
[0037] FIG. 1 is a conceptual diagram illustrating a composite AIoT-based industrial site worker safety monitoring system according to one embodiment of the present invention.
[0038] FIG. 2 is a conceptual diagram for explaining a field terminal unit according to one embodiment.
[0039] FIG. 3 is a block diagram illustrating a terminal sensing unit according to one embodiment.
[0040] FIG. 4 is a block concept diagram for explaining a smart video capturing unit according to one embodiment.
[0041] FIG. 5 is a block diagram illustrating a thermal imaging unit according to one embodiment.
[0042] FIG. 6 is a block diagram illustrating a terminal integration analysis unit according to one embodiment.
[0043] As illustrated in FIGS. 1 to 6, the composite AIoT-based industrial site worker safety monitoring system according to the present embodiment includes a field terminal unit (100) that uses various IoT sensors and equipment to sense and detect safety at the site and analyzes the occurrence of risks based on this, a server unit (200) that monitors the analysis results of the field terminal unit (100), and a mobile terminal unit (300) that receives the notification results of the server unit (200).
[0044] Additionally, it includes a field air conditioning unit (400) that controls the field environment.
[0045] The complex AIoT (Artificial Intelligence of Things)-based industrial site worker safety monitoring system presented in this example ensures the safety of industrial site workers and can proactively prevent human accidents by detecting abnormalities, dangerous situations, and exposure to hazardous environments in real time through integrated analysis of sensing data linked with on-site sensors and CCTVs. Furthermore, by comprehensively linking various work environment sensors with acoustic and thermal detection equipment and analyzing the IoT information input from them using AI, the system can rapidly detect anomalies or risks on-site and enable on-site response measures. Additionally, it can transmit situational information to a server in real time, allowing managers to take appropriate action.
[0046] The field terminal (100) effectively collects and stores various safety-related data generated at the industrial site, namely the state of workers and the work environment, in real time through various IoT sensors and equipment, integrates the collected data of various types and heterogeneous types, and comprehensively determines the dangerous situation by performing complex analysis based on AI. In this example, the field terminal (100) is effective as an edge computing-based safety judgment unit that integrates sensing, analysis, communication, and control functions into a single embedded system.
[0047] The field terminal unit (100) includes a terminal sensing unit (110) that senses the industrial site, a terminal integrated analysis unit (120) that analyzes whether a risk occurs based on the sensed information, a terminal communication unit (130) that provides analysis results and receives an external control signal through communication with a server unit (200), and a terminal control unit (140) that controls the operation of each unit according to the control signal.
[0048] The terminal sensing unit (110) can simultaneously collect different physical quantities such as environment, image, heat distribution, and sound to detect complex risk situations that are difficult to identify with a single sensor. At this time, the data from each sensor is collected based on a common timestamp, enabling mutual correlation analysis during the analysis phase.
[0049] The terminal sensing unit (110) includes an environment sensor unit (111) that senses environmental information of the site, a smart video capturing unit (112) that captures the site conditions and the actions of the worker, a thermal imaging capturing unit (113) that captures the heat distribution of the site and the worker, and an acoustic sensor unit (114) that detects abnormal sounds and danger signals.
[0050] The environmental sensor unit (111) is effective in quantifying hazardous environmental elements of the industrial site's workspace and detecting hazardous environments.
[0051] It is effective for the environmental sensor unit (111) to use sensors to detect temperature, humidity, fine dust, TVOC, CO, CO2, combustible gas (LPG / LNG) and formaldehyde, etc., in order to monitor indoor air quality and safety.
[0052] The environmental sensor unit (111) can use temperature, humidity, fine dust and TVOC multi-sensors, CO2 sensor, carbon monoxide sensor, LPG and LNG gas sensor and formaldehyde sensor.
[0053] Multi-sensors for temperature, humidity, fine dust, and TVOC can utilize sensors that integrate temperature, humidity, fine dust, and TVOC into a single module. Sensors can be used with a structure combining a digital IC sensor for temperature and humidity, a laser / infrared optical dust sensor for fine dust, and a semiconductor or VOC IC sensor for TVOC. For CO2 sensors, non-dispersive infrared sensors that measure CO2 concentration by absorbing infrared rays of specific wavelengths can be used. Carbon monoxide sensors utilize electrochemical or semiconductor types, and it is possible to measure concentrations in ppm units through the principle of generating current proportional to concentration. LPG / LNG gas sensors use metal oxide semiconductor methods to determine concentration based on the characteristic change in resistance when gas is adsorbed onto the sensor surface. Formaldehyde sensors use electrochemical or semiconductor gas sensors to measure ug / m³ 3 Alternatively, it is possible to measure the concentration in PPM units.
[0054] It is effective for the smart video recording unit (112) to analyze the worker's actions and the situation at the site based on video. The smart video recording unit (112) may use a CCTV or an AI camera. It is effective for the smart video recording unit (112) to acquire video data including the worker's location, movement, and posture information, as well as whether protective gear is being worn and whether the worker has approached a dangerous area.
[0055] The smart video recording unit (112) includes a camera unit (112a) for recording video, an actuation unit (112b) for moving the camera unit (112a) according to a video processing module unit (112c), a video processing module unit (112c) for detecting a worker in the recorded video and determining the movement of the worker to track the worker through the camera unit (112a), and a video storage module (112d) for storing the video recorded by the camera unit (112a).
[0056] The camera unit (112a) can use a standard camera or CCTV capable of capturing images, and it is possible to use a camera that supports PTZ.
[0057] It is preferable that the actuation unit (112b) includes a motor, a driver, and an encoder for moving the camera unit (112a). It is effective for the actuation unit (112b) to operate according to a control signal of the image processing module (112c) to move the camera unit (112a) in a target direction.
[0058] The image processing module (112c) receives an image captured through the camera unit (112a), detects a worker within the image, analyzes the movement direction of the detected worker, and can track and capture the worker through the camera unit (112a).
[0059] It is effective for the image processing module (112c) to maintain the worker ID and center coordinates for each image frame through deep learning-based human detection technology and multi-object tracking function. It is possible to continuously track the worker by calculating the movement vector, direction, and speed based on the difference in center coordinates between the current frame and the next frame, and moving the camera to position the worker in the image at the center of the image based on the result. In other words, it is possible to maintain the worker's center coordinates for each frame. To achieve this, it is desirable to apply a moving average or a prediction filter. To this end, it is effective for the image processing module (112c) to output a control signal to the actuation unit (112b) to move the camera unit (112a) along the calculated worker movement direction. Of course, it is also possible to adjust the movement direction and speed of the camera unit (112a) through the error between the center of the screen and the worker's center coordinates.
[0060] The above-described image processing module (112c) can perform its function in the terminal integrated analysis unit (120). Since the field terminal (100) of this example is manufactured in the form of a single embedded module, that is, a board, similar functions can be integrated. To this end, information stored in the image storage unit (112d) is provided to the terminal integrated analysis unit (120), and it is effective for the terminal integrated analysis unit (120) to perform the function of the image processing module (112c). Since the terminal integrated analysis unit (120) and the image storage unit (112d) are embedded within a single board and the camera unit (112a) is electrically connected to this board, the image of the camera unit (112a) is stored in the image storage unit (112d) in real time, and the terminal integrated analysis unit (120) analyzes this in real time and provides a control signal to the actuation unit (112b). Since it is a single board, there is no time delay, and high-speed operation and function implementation become possible.
[0061] It is effective for the video storage unit (112d) to store the video captured through the camera unit (112a) in real time. It is desirable for the video storage unit (112d) to store the stored video for a short period. Through this, it is possible to manage the amount of data and protect personal information, thereby ensuring smooth operation of the system. Of course, if necessary, the video storage unit (112d) can store the video stored in real time and the video stored for a certain period using different filters. That is, the video stored in real time may not require separate editing so that it can be analyzed in real time. However, for the video that has been stored for a certain period, it is effective to protect personal information by applying a filter such as blurring the face area. Of course, it is effective not to use such filters, but to store only video information when an abnormality or danger occurs according to the control signal of the terminal integrated analysis unit (120), terminal control unit (140), or server unit (200), or to store only certain frame information instead of video information.
[0062] The thermal imaging unit (113) can measure the heat distribution of the site or the worker in a non-contact and non-identification manner. The thermal imaging unit (113) uses an infrared sensor to acquire the surface temperature distribution of the worker, equipment, and workspace in the form of an image, and through this, can determine the worker's high-temperature condition, the risk of heat stress, and whether the worker is approaching high-temperature equipment. In addition, it is possible to determine the worker's location, movement, and posture information, as well as whether protective gear is being worn, separately from the smart image capturing unit (112) mentioned above.
[0063] The thermal imaging unit (113) may include a thermal imaging camera unit (113a) for capturing a thermal image, a thermal imaging actuation unit (113b) for moving the thermal imaging camera unit (113a), a thermal imaging calculation module unit (113c) for detecting a worker within the thermal imaging information and determining the movement of the detected worker to control the movement of the thermal imaging camera unit (113a), and a thermal imaging storage unit (113d) for storing thermal imaging information captured by the thermal imaging camera unit (113a).
[0064] It is effective for the thermal imaging camera unit (113a) to capture images using an infrared sensor. Since thermal images are captured using the thermal imaging camera unit (113a) of this example, the exposure of personal identification information is limited, which can be advantageous for protecting personal information.
[0065] It is preferable that the thermal imaging actuation unit (113b) includes a motor, a driver, and an encoder for moving the thermal imaging camera unit (113a). It is effective for the thermal imaging actuation unit (113b) to operate according to a control signal of the thermal imaging calculation module unit (113c) to move the thermal imaging camera unit (113a) in a target direction.
[0066] The thermal image calculation module (113c) receives thermal image information captured through the thermal image camera unit (113a), detects a worker within the image information, analyzes the movement, posture, and direction of movement of the detected worker, and enables tracking and capturing the worker through the thermal image camera unit (113a). Additionally, it is possible to detect a point or area where the temperature changes rapidly within the thermal image and move the thermal image camera unit (113a) to the detected area to monitor that area.
[0067] The thermal image calculation module (113c) effectively distinguishes workers or areas with high thermal changes per thermal image frame and maintains the center coordinates of the image through deep learning-based human detection technology and multi-object tracking function. The movement vector, direction, and velocity can be calculated based on the difference in center coordinates between the current frame and the next frame. Additionally, thermal changes between the two frames can be calculated by comparing them. Based on the results, it is possible to continuously track the workers in the thermal image by moving the thermal camera so that the center coordinates of the image or the area with high thermal changes are in the center of the image.
[0068] It is possible to maintain the center coordinates of the worker or the area with significant thermal change on a frame-by-frame basis. To achieve this, it is desirable to apply a moving average or a prediction filter. To this end, the thermal image calculation module (113c) effectively outputs a control signal to the thermal image actuation unit (113b) to move the thermal image camera unit (113a) along the calculated direction of movement of the worker or to move the thermal image camera unit (113a) according to the direction of thermal change. Of course, it is also possible to adjust the direction and speed of movement of the thermal image camera unit (113a) through the error between the center of the screen and the center coordinates of the worker.
[0069] The above-described thermal image calculation module (113c) can perform its function in the terminal integrated analysis unit (120). Since the field terminal (100) of this example is manufactured in the form of a single embedded module, that is, a board, similar functions can be integrated. To this end, information stored in the thermal image storage unit (113d) is provided to the terminal integrated analysis unit (120), and it is effective for the terminal integrated analysis unit (120) to perform the function of the thermal image calculation module (113c). Since the terminal integrated analysis unit (120) and the thermal image storage unit (113d) are embedded within a single board and the thermal image camera unit (113a) is electrically connected to this board, the image of the thermal image camera unit (113a) is stored in the thermal image storage unit in real time, and the terminal integrated analysis unit (120) analyzes this in real time and provides a control signal to the thermal image actuation unit (113b). Since it is a single board, there is no time delay, and high-speed operation and function implementation become possible.
[0070] It is effective for the thermal image storage unit (113d) to store images captured through the thermal image camera unit (113a) in real time. It is effective for the thermal image storage unit (113d) to store the captured images for a longer period than the storage period in the image storage unit. This is because the images stored in the thermal image storage unit (113d) have no possibility of personal information being exposed, and based on thermal changes, it is possible to quickly determine not only the actions of the worker but also whether safety equipment is being worn and whether a high-temperature environment is occurring, and it is also possible to store and convert this into dB.
[0071] The acoustic sensor unit (114) is effective at detecting abnormal sounds and danger signals occurring in industrial sites.
[0072] The acoustic sensor unit (114) may use a microphone or a microphone array. It is effective to have two or more acoustic sensors for detecting sound in a 360-degree direction of the installation area. At this time, it is desirable for the installed sensors to have the same spacing. For example, if there are two sensors, it is desirable to have a spacing of 180 degrees, and if there are four sensors, it is desirable to have a spacing of 90 degrees each. Through this, it is possible to detect sound in a 360-degree direction of the installation site. It is effective for the acoustic sensors to remove background noise and extract specific dangerous sound patterns through digital signal processing (DSP).
[0073] The acoustic sensor unit (114) may be equipped with a module for analyzing abnormal sound patterns. It is effective to identify specific acoustic patterns related to dangerous situations, such as screams, explosions, collisions, and abnormal machine noises, in real time using acoustic event detection technology. Of course, acoustic information collected by the acoustic sensor unit (114) can be provided to the terminal integrated analysis unit to perform acoustic frequency conversion and feature extraction, and can be used in combination with video and other environmental sensing information for risk assessment.
[0074] It is effective for the terminal integrated analysis unit (120) to individually analyze multi-sensor data provided from the terminal sensing unit (110) and fuse the results to make a final determination of abnormal and dangerous situations at the site. It is desirable for the terminal integrated analysis unit (120) to perform edge AI analysis using edge computing. It is effective for the terminal integrated analysis unit (120) to independently analyze the environment, video, thermal image, and sound using optimized analysis modules, and then make an integrated determination of the analysis results. Through this, it is possible to efficiently distribute the computational load in an embedded environment, and to facilitate the lightweighting, replacement, and updating of individual modules.
[0075] The terminal integrated analysis unit (120) includes an environment analysis module (121) that analyzes the environmental conditions of the site, an image analysis module (122) that analyzes captured images, a thermal image analysis module (123) that analyzes thermal images, an acoustic analysis module (124) that analyzes acoustics, and an integrated judgment unit (125) that determines whether an abnormality or danger occurs based on the analysis results of each analysis module.
[0076] Additionally, the terminal integration analysis unit (120) may further include a learning correction module and a privacy protection module.
[0077] The environmental analysis module (121) can perform threshold determination and abnormal pattern analysis based on the provided temperature, humidity, gas, and fine dust sensing information. That is, it is possible to analyze not only whether a simple threshold, i.e., a reference value is exceeded, but also the rate of increase, duration, and whether complex conditions are satisfied. Through this, it is possible to derive the risk of heat stress, the possibility of gas leakage, and poor ventilation conditions, and the analysis results can be provided to the integrated judgment unit (125) in the form of a risk index or analysis results.
[0078] The environmental analysis module (121) can analyze the risk of exposure to a hazardous environment in this way and, based on this, generate a control signal for air conditioning control, that is, for controlling the field air conditioning unit (400).
[0079] The environment analysis module (121) can learn patterns of environmental data, i.e., sensing information, using a lightweight LSTM or GRU model and predict sudden or abnormal changes in real time. Since environmental data has low dimensionality, it is effective to minimize the complexity of the model and maximize the inference speed through quantization.
[0080] The video analysis module (122) can analyze the worker based on input video information and analyze the dangerous situation at the site. The video analysis module (122) can recognize and track the worker, analyze the worker's actions, and additionally analyze whether the worker is wearing safety equipment.
[0081] The video analysis module (122) can detect a worker as an object in a general video and analyze the worker's movement path and dwell area through multi-object tracking (MOT). In addition, it is possible to recognize behaviors such as falling, abnormal posture, and approaching a dangerous area (leaving a designated area) through a pose estimation and behavior classification model. Furthermore, it can determine and analyze whether safety equipment such as a safety helmet, mask, and protective suit is being worn.
[0082] Considering the embedded environment, it is also possible to apply lightweight CNN-based models such as MobileNet and YOLO-Nano, as well as frame skipping and RoI-based inference techniques. By defining knowledge from large models in lightweight models and removing unnecessary connections, the size and computational load of the models can be reduced, enabling real-time inference even at low FPS.
[0083] The video analysis module (122) can detect a moving worker, check the direction of movement of the worker's center coordinates in the current video frame and the next video frame, and move the camera unit toward the worker to monitor safety accidents of the worker.
[0084] The video analysis module (122) can detect, based on deep learning, whether a worker is wearing safety equipment such as a safety helmet, safety shoes, dust / gas mask, and safety glasses at the site (workplace). Although the use of high-performance GPUs and cloud computers is required for deep learning analysis, in this example, it is effective to analyze the video in a short time, almost in real-time, in an embedded edge environment. It is desirable to perform a lightweight deep learning algorithm and hardware optimization. To perform deep learning, it passes through three hidden layers. That is, it includes Low Level Features, Mid Level Features, and High Level Features. It is then desirable to output after a Classifier.
[0085] The video analysis module (122) effectively recognizes motion in real time by using morphological information of the Motion History Image (MHI) for motion analysis. That is, when a video is input, the MHI is extracted, the motion occurrence area is detected, the gradient image is extracted, shape information with applied shape context is extracted, and the motion is recognized from this. Through this, it becomes possible to detect falling (falling over), etc.
[0086] For the video analysis module (122), a deep learning model using GPU-based computing resources can be used, and it is effective to maintain an accurate learning model while keeping it small in size and simplifying computations. In other words, efficiency can be increased compared to existing models through lightweight deep learning technology. To achieve this, it is effective to use a model compression technique that reduces the parameters of the existing model to lighten the algorithm. That is, lightweighting can be achieved by applying a technique through quantization. Through this, the number of model parameters can be reduced, but it is also possible to perform lightweighting by reducing the representation size of the parameters. By applying multi-core processing, core computations within the CPU are distributed to perform deep learning computations, thereby obtaining faster execution speeds.
[0087] The thermal imaging analysis module (123) is capable of non-contact body temperature measurement, object recognition and behavior analysis based on heat distribution, and analysis of whether safety equipment is worn, as well as the spread of thermal hazards. It can perform thermal hazard and worker condition analysis based on the heat distribution of the worker and the site.
[0088] The thermal image analysis module (123) can separate a worker into a thermal object based on infrared image information and analyze temperature distribution and changes over time. Through this, it is possible to determine whether the worker is in a high-temperature state, whether heat stress has accumulated, and whether they have accessed high-temperature equipment, and stable worker tracking is possible even in environments with lighting changes, smoke, or reduced visibility. In addition, since color and shape information is displayed in a limited manner unlike image information, the possibility of personal identification is low, which is advantageous for protecting personal information. Furthermore, it is possible to indirectly determine whether protective equipment is being worn by utilizing the difference in heat generated when wearing and not wearing protective equipment, and it is also possible to analyze the movement and abnormal behavior patterns of the thermal object, which is the worker.
[0089] The thermal image analysis module (123) can define a worker from the detected heat information and analyze the maximum temperature of a specific area of the worker. Through this, it is possible to detect and analyze high-temperature conditions and increase the measurement accuracy of the heat source through emissivity correction and distance correction algorithms. Since the thermal image information has little background noise and the human shape can be clearly distinguished, it is possible to extract and track worker objects through semantic segmentation or heatmap-based clustering. In addition, when analyzing abnormal movements such as falling, it may not be affected by changes in the surrounding environment, such as lighting. Furthermore, the thermal image analysis module (123) can detect and analyze abnormal high-temperature sources in the surrounding environment, such as overheated machinery or early signs of fire.
[0090] The thermal image analysis module (123) is preferably lightweighted by reducing the depth and width of the CNN to use a low resolution and a small number of channels for lightweighting.
[0091] The thermal image analysis module (123) detects a moving worker through the thermal image camera unit (113a), checks the movement of the worker's center coordinates in the current image frame and the next image frame to analyze the direction of movement, and moves the camera toward the worker to monitor and determine the worker's safety accident.
[0092] The acoustic analysis module (124) is effective in analyzing abnormal sounds that cause accidents and dangers based on the provided acoustic information. It can collect acoustic signals associated with safety accidents, such as collision sounds, explosion sounds, screams, and abnormal machine sounds, and classify whether the sound is dangerous by extracting spectrum-based features. In addition, it can quickly detect sudden events occurring in a short period of time, thereby serving as a pre-warning mechanism rather than video analysis. Furthermore, by additionally reflecting the acoustic analysis results through the integrated judgment unit, it can determine whether a danger has occurred more accurately.
[0093] The acoustic analysis module (124) detects acoustic events. It converts acoustic, or audio information, into a Mel spectrogram and then uses a lightweight CNN model to classify screams, explosions, collisions, and abnormal sounds of specific machines in real time. By applying preprocessing filtering for on-site noise removal, the accuracy of the analysis can be increased, and by simplifying the structure of the model, continuous monitoring with low power consumption can be enabled. It is possible to acquire abnormal sounds generated in the surrounding 360-degree environment and to collect and analyze acoustic information from places or spaces that cannot be visually, or videoly verified.
[0094] It is desirable for the integrated judgment unit (125) to fuse the results of each analysis module unit to make a final determination of whether actual risk has occurred.
[0095] The integrated judgment unit (125) can combine environmental, video, thermal, and acoustic analysis results and determine whether there is a risk by considering the reliability of the results and the timing of occurrence. For example, if the video analysis result indicates that a worker has fallen, the risk is determined based on the analysis results of other modules. That is, if the thermal analysis result indicates high heat and the acoustic analysis result indicates an abnormal fall rather than a normal fall, it is possible to determine whether there is a risk by raising the risk level. Of course, each module may have a logic that immediately determines the risk if the analysis result exceeds the risk threshold value.
[0096] The integrated judgment unit (125) can receive the analysis results of each analysis module as risk scores or classification results, comprehensively judge the final risk situation, and determine the response step.
[0097] The integrated judgment unit (125) can perform data fusion and situation recognition simultaneously. That is, through the integration of decision levels, it is possible to calculate a final composite risk index by synthesizing the results of surrounding environment analysis, image analysis, thermal image analysis, and acoustic analysis. To this end, the reliability of judgment regarding composite risk situations can be improved by using a roll-based expert system or fuzzy logic. It is possible to determine a response level based on such judgment results and provide the result to the terminal control unit.
[0098] As described in this example, general video information and thermal image information are analyzed separately, and an integrated judgment can be made through an integrated judgment unit. This enhances reliability through mutual complementarity, ensures both privacy and functionality simultaneously, and enables accurate judgment of complex risk situations while reducing false positives. In other words, thermal images can compensate for the weaknesses of general video, which is susceptible to lighting, shadows, and obstructed views. By using thermal images for tracking primary workers and detecting body temperature to minimize the exposure of personally identifiable information, and by allowing general video to analyze only specific areas requiring high-resolution information, such as the detailed wearing status of safety equipment, functionality and privacy protection can be satisfied simultaneously. Furthermore, even if one sensor makes a false positive, if another sensor determines the situation to be normal, the issuance of an alarm—that is, the judgment of danger occurrence—is withheld, thereby reducing unnecessary alarms. Conversely, if one sensor misses a risk, another sensor can detect it, thereby improving the actual risk detection rate.
[0099] The learning information correction unit can provide information provided to the terminal integrated analysis unit (120) and correct the learning information to be used in each analysis module based on this. In addition, it is possible to receive additional external analysis information.
[0100] In addition, it is desirable to de-identify personally identifiable information within the image information stored in the learning information correction unit through a privacy protection module.
[0101] It is effective for the terminal communication unit (130) to transmit the results analyzed through the terminal integrated analysis unit (120) to the server unit (200) and to receive external control signals. The terminal communication unit (130) can perform bidirectional communication with the server unit via wired / wireless communication. It can provide analysis results, status information, and risk judgment information. Additionally, it is possible to receive external control signals from the server unit (200), such as air conditioning control or system setting changes.
[0102] It is effective for the terminal control unit (140) to notify a separate risk based on the analysis result of the terminal integration analysis unit (120) and to control the operation of each unit according to an external control signal.
[0103] The terminal control unit (140) is equipped with a separate alarm unit or notification unit, and when a dangerous situation is detected through the terminal integrated analysis unit (120), it is possible to notify the industrial site in real time and also provide local notification to the surrounding area. The operation of each unit can be managed in an integrated manner through sensor driving, analysis cycle control, communication control, and air conditioning linkage signal output. In addition, it is desirable to be able to respond in conjunction with the air conditioning unit when a danger occurs.
[0104] It is effective for the server unit (200) to monitor the risk analysis results transmitted from the field terminal unit (100) and to propagate the results through the manager and the mobile terminal unit (300).
[0105] The server unit (200) stores information provided by the field terminal unit (100) and visualizes it in the form of a central dashboard to provide warning notifications to the administrator. It is effective for the server unit (200) to store and manage a large amount of information in a time-series manner through stable connections with multiple field terminal units (100). Depending on the large amount of data and the level of real-time response requirements, a cloud server can be established and used as training data for the terminal integrated analysis unit (120) through immediate on-site response and long-term data analysis. That is, it is possible to manage training data for accident pattern analysis, risk statistics, and AI model improvement through historical data analysis.
[0106] The mobile terminal unit (300) can receive dangerous situations through the server unit (200) and notify the terminal user of them. That is, it is effective to receive and provide danger alerts, status information, video and thermal image snapshots from the server unit (200). This can induce immediate response and decision-making regarding the scene. In addition to a smartphone, it is also possible to use a smart watch or band as the mobile terminal unit (300).
[0107] It is effective for the on-site air conditioning unit (400) to be equipped with various devices and facilities for controlling the on-site environment. It is also possible to actively control the on-site environment to mitigate dangerous situations when they occur. The on-site air conditioning unit (400) can control high temperatures, harmful gases, and deteriorating air quality through ventilation, exhaust, and heating and cooling.
[0108] It is effective for the field air conditioning unit (400) to include a heating and cooling module, an air purifier, a ventilation fan, and a dehumidifier. It is effective for this field air conditioning unit (400) to be electrically connected to the field terminal unit (100). Through this, it is possible to operate immediately according to the analysis results of the terminal integration analysis unit (120) of the field terminal unit (100) to quickly mitigate dangerous situations. In this way, various equipment and facilities are connected in the form of an embedded board, and control is performed through an edge device to quickly detect abnormalities or dangers at the site and quickly improve dangerous situations.
[0110] The present invention is not limited to the embodiments described above, and it is possible to construct an effective composite AIoT-based industrial site worker safety monitoring system through various modification technologies. Below, a composite AIoT-based industrial site worker safety monitoring system according to various modifications of the present invention is described. Descriptions that overlap with the embodiments described above are omitted in the following description. Furthermore, the technology described below may be applied to the embodiments described above.
[0111] FIG. 7 is a conceptual diagram illustrating a composite AIoT-based industrial site worker safety monitoring system according to one variant of the present invention.
[0112] FIG. 8 is a block diagram illustrating an integrated judgment unit according to one variant example.
[0113] FIG. 9 is a block diagram illustrating an equipment analysis section according to one modified example.
[0114] FIG. 10 is a block diagram illustrating a field analysis section according to one variant example.
[0115] As illustrated in FIGS. 7 to 10, the composite AIoT-based industrial site worker safety monitoring system according to the present variation includes an equipment detection unit (500) that detects equipment abnormalities installed at an industrial site, a site detection unit (600) that detects workers and the environment at the industrial site, an integrated judgment unit (700) that determines whether a risk has occurred using the detection results of the equipment detection unit (500) and the site detection unit (600), a server unit (800) that monitors information from the integrated judgment unit (700), and a mobile terminal unit (900) that receives notification results from the server unit (800).
[0116] Of course, it may further include a field air conditioning unit connected to the integrated judgment unit (700) and managing the environment of the industrial site.
[0117] Through this, the complex AIoT-based industrial site worker safety monitoring system of the present modification can analyze and determine matters where worker safety is threatened due to equipment malfunctions, and can reduce the occurrence of risks caused by equipment by performing predictive maintenance on the equipment.
[0118] The equipment detection unit (500) is effective in sensing and analyzing mechanical, thermal, electrical, and acoustic abnormal signals generated from industrial equipment from various angles. Through this, it is desirable to detect abnormal conditions and signs of failure of the equipment at an early stage and to make a judgment through the integrated judgment unit.
[0119] It is effective for the equipment detection unit (500) to include various sensors that are attached to or in close contact with the equipment to detect mechanical, thermal, and electrical characteristics generated by the equipment. The equipment detection unit (500) may include a vibration sensor that detects abnormal vibrations of the equipment. That is, it is effective to detect shocks or abnormal operation caused by bearing wear, shaft imbalance, or misalignment. In addition, it may detect changes in motor load, overload, or signs of insulation deterioration through current or voltage sensors. It may detect heat generation of the equipment through a temperature sensor, and detect friction sounds, metal collision sounds, or mechanical abnormal noises of the equipment through an acoustic sensor. Furthermore, it may detect RPM instability, slip, or idle operation through rotation or speed sensors.
[0120] In this example, the sensor of the equipment detection unit (500) also uses an IoT sensor, and it is effective for each sensor to be electrically connected to an embedded board, that is, an integrated judgment unit (700) which is an integrated board, to provide the detection signal.
[0121] The field detection unit (600) may be equipped with an environment sensor unit, a smart video capturing unit, a thermal imaging capturing unit, and an acoustic sensor unit as described in the previous example.
[0122] Through this, it is possible to provide surrounding environment information, image information, thermal image information, and acoustic information to the integrated judgment unit.
[0123] The integrated judgment unit (700) may include an equipment analysis unit (710) that analyzes equipment abnormalities based on the detection results of the equipment detection unit (500), a field analysis unit (720) that analyzes worker and environmental abnormalities based on the detection results of the field detection unit (600), an integrated analysis unit (730) that analyzes and judges dangerous situations based on the analysis results of the equipment analysis unit (500) and the field analysis unit (600), an integrated control unit (740) that controls the operation of each unit, and an integrated communication unit (750) that communicates with a server.
[0124] The equipment analysis unit (710) includes a vibration analysis module (711) for analyzing vibration abnormalities of the equipment, an electrical analysis module (712) for analyzing electrical characteristic abnormalities of the equipment, a temperature analysis module (713) for analyzing temperature abnormalities of the equipment, an acoustic analysis module (714) for analyzing acoustic abnormalities of the equipment, and a deterioration analysis module (715) for analyzing whether the equipment is deteriorated.
[0125] The vibration analysis module (711) can convert machine vibration data from the time domain to the frequency domain and analyze specific frequencies generated in a rotating body or bearing to analyze abnormal peak occurrences. It is also possible to analyze abnormal patterns by extracting amplitude fluctuations of the signal. At this time, it is also possible to analyze rapid increases in vibration through the difference between the average level and the maximum amplitude of vibration energy.
[0126] The electrical analysis module (712) can analyze the occurrence of electrical abnormalities through the analysis of current waveform distortion and frequency fluctuations. For example, when the input current of a motor is normal, it shows a constant pattern, but when an abnormality occurs, an abnormal waveform is generated. In addition, it can detect minute changes in power frequency and detect the fluctuation range to detect abnormalities early.
[0127] The temperature analysis module (713) is effective for analyzing the absolute temperature, its rate of rise, and the deviation from normal. That is, it can measure the absolute temperature of the equipment, evaluate the rate of rise over a certain period of time, and analyze whether there is an abnormality by determining how statistically the current temperature deviates from past normal operation data.
[0128] The acoustic analysis module (714) can perform energy analysis by frequency band and classify dangerous sound patterns.
[0129] In addition, the deterioration analysis module (715) manages the trend of changes in the state of the equipment over time based on the results analyzed by each analysis module, and based on this, can predict planned maintenance or equipment failure. It is effective to manage the analysis of abnormalities in each part in the form of a separate score accumulation and to predict the rate of deterioration and the timing of maintenance.
[0130] The field analysis unit (720) may include an environment analysis module (721) that analyzes the environmental conditions of the site, an image analysis module (722) that analyzes captured images, a thermal image analysis module (723) that analyzes thermal images, and an acoustic analysis module (724) that analyzes acoustics.
[0131] The environmental analysis module (721) can perform threshold determination and abnormal pattern analysis based on the provided temperature, humidity, gas, and fine dust sensing information. The image analysis module (722) can detect a worker as an object in a general image and analyze the worker's movement path and dwell area through multi-object tracking (MOT). The thermal image analysis module (723) is capable of non-contact body temperature measurement, object recognition and behavior analysis based on heat distribution, as well as analyzing whether safety equipment is worn and the spread of thermal hazards. The acoustic analysis module (724) is effective in analyzing abnormal sounds that cause accidents and dangers based on the provided acoustic information.
[0132] The integrated analysis unit (730) can combine the equipment analysis results from the equipment analysis unit (710) with the site and worker analysis results from the site analysis unit (720) and determine whether there is a risk by considering the reliability of the results and the timing of occurrence. For example, if the vibration and heat analysis of the equipment shows abnormalities, and the analysis shows a worker collapsing and abnormal gas generation, it determines that a risk has occurred due to the equipment abnormality. Through this, the risk level can be raised, and the occurrence of a risk can be notified to take measures to move the relevant equipment and surrounding workers. Of course, each module unit may have a logic that immediately determines a risk if the analysis results exceed a risk threshold value.
[0133] The integrated analysis unit (730) can receive the analysis results of each analysis module as risk scores or classification results, comprehensively determine the final risk situation, and determine the response level. The integrated analysis unit (730) can perform data fusion and situational awareness simultaneously. That is, through the integration of decision levels, it is possible to calculate a final composite risk index by synthesizing the results of equipment anomaly analysis, surrounding environment analysis, image analysis, thermal image analysis, and acoustic analysis. It is possible to determine the response level based on such judgment results and provide the result to the integrated control unit.
[0134] It is effective for the integrated control unit (740) to notify of separate risks based on the analysis results of the equipment analysis unit (710), the field analysis unit (720), and the integrated analysis unit (730), and to control the operation of each unit according to an external control signal. The integrated control unit (740) may be equipped with a separate alarm unit or notification unit to notify the industrial site in real time when a dangerous situation is detected, and to provide a local notification to the surrounding area.
[0135] It is effective for the integrated communication unit (750) to transmit the analyzed results of the equipment analysis unit (710), the field analysis unit (720), and the integrated analysis unit (730) to the server unit (800) and to receive external control signals. The integrated communication unit can perform bidirectional communication with the server unit (800) via wired / wireless communication.
[0136] It is effective for the server unit (800) to monitor the risk analysis results transmitted from the equipment detection unit (710), the site detection unit (720), and the integrated judgment unit (730), and to propagate the results through the manager and the mobile terminal unit (900).
[0137] The mobile terminal unit (900) receives a dangerous situation through the server unit (800) and can notify the terminal user of this.
[0138] Although the technical concept of the present invention described above has been specifically described in preferred embodiments, it should be noted that the aforementioned embodiments are for illustrative purposes only and are not intended to be limiting. Furthermore, a person skilled in the art will understand that various embodiments are possible within the scope of the technical concept of the present invention. Explanation of the symbols
[0140] 100: Field terminal unit 110: Terminal sensing unit 111: Environmental sensor unit 112: Smart video recording unit 113: Thermal imaging unit 114: Acoustic sensor unit 120: Terminal Integrated Analysis Unit 121: Environment Analysis Module Unit 122: Video Analysis Module 123: Thermal Imaging Analysis Module 124: Acoustic Analysis Module 125: Integrated Judgment Unit 130: Terminal communication unit 140: Terminal control unit 200, 800: Server unit 300, 900: Mobile terminal unit 400: On-site HVAC unit 500: Equipment detection unit 600: On-site detection unit 700: Integrated judgment unit 710: Equipment Analysis Department 720: Field Analysis Department 730: Integrated Analysis Unit 740: Integrated Control Unit 750: Integrated Communications Department
Claims
Claim 1 A field terminal unit that senses and detects on-site safety using various IoT sensors and equipment, and analyzes the occurrence of risks based on this; a server unit that monitors the analysis results of the field terminal unit; The system includes a mobile terminal that receives notification results from a server unit, and the field terminal includes a terminal sensing unit that senses an industrial site, a terminal integrated analysis unit that analyzes whether a risk has occurred based on the sensed information, a terminal communication unit that provides analysis results to the server unit and receives external control signals, and a terminal control unit that controls the operation of each unit according to the control signals. The terminal integrated analysis unit includes an environment analysis module that analyzes the environmental conditions of the site, an image analysis module that analyzes captured images, a thermal image analysis module that analyzes thermal images, and an acoustic analysis module that analyzes acoustics. It also includes an integrated judgment unit that determines whether an anomaly or risk has occurred based on the analysis results of each analysis module, a learning information correction unit that provides information to the terminal integrated analysis unit and corrects learning information to be used by each analysis module based on this information, and a privacy protection module that de-identifies information capable of personal identification within the image information stored in the learning information correction unit. The integrated judgment unit performs data fusion and situational awareness simultaneously, calculates a final composite risk index by synthesizing the results of surrounding environment analysis, image analysis, thermal image analysis, and acoustic analysis, and determines the final risk situation and response considering reliability and the timing of occurrence. A complex AIoT-based industrial site worker safety monitoring system characterized by determining a stage, wherein the terminal sensing unit includes an acoustic sensor unit that detects abnormal sounds and danger signals, wherein the acoustic sensor unit is equipped with two or more acoustic sensors for 360-degree acoustic detection, wherein the acoustic sensors remove background noise and extract specific danger sound patterns through digital signal processing (DSP), and wherein the acoustic sensor unit is equipped with a module that analyzes abnormal sound patterns. Claim 2 delete Claim 3 A composite AIoT-based industrial site worker safety monitoring system according to claim 1, wherein the terminal sensing unit further comprises an environment sensor unit that senses environmental information of the site, a smart video capturing unit that captures the site situation and the actions of the worker, and a thermal imaging unit that captures the heat distribution of the site and the worker. Claim 4 In claim 3, the environmental sensor unit uses temperature, humidity, fine dust and TVOC multi-sensors, CO2 sensor, carbon monoxide sensor, LPG and LNG gas sensor, and formaldehyde sensor; the smart image capturing unit includes a camera unit that captures images, an actuation unit that moves the camera unit according to an image processing module, an image processing module that detects a worker in the captured images and determines the movement of the worker to track the worker through the camera unit, and an image storage module that stores images captured by the camera unit; and the thermal imaging unit includes a thermal imaging camera unit that captures thermal images, a thermal imaging actuation unit that moves the thermal imaging camera unit, a thermal imaging processing module that detects a worker in the thermal imaging information and determines the movement of the detected worker to control the movement of the thermal imaging camera unit, and a thermal imaging storage unit that stores thermal imaging information captured by the thermal imaging camera unit, thereby forming a complex AIoT-based industrial site worker safety monitoring system. Claim 5 delete Claim 6 delete
Citation Information
Patent Citations
Edge Computing and Cloud-Based Factory Energy Management System using big-data
KR102535230B1
Control system for preventing safety accidents and operation method thereof
KR1020230051851A
Integrated safety managementing system for industrial site and method thereof
KR1020240070749A
Ai and IoT based safety management system
KR1020250149873A
