A smart safety protection system for ship construction

Abstract

This invention discloses an intelligent safety protection system for shipbuilding, comprising: wearable intelligent safety detection equipment, a work environment detection and personnel management system, an AI perception prediction and evaluation module, and an integrated safety production detection platform. The wearable intelligent safety detection equipment and the work environment detection and personnel management system collect multi-source data, which is then input into the AI ​​perception prediction and evaluation module. The AI ​​perception prediction and evaluation module generates risk types and risk levels through cross-modal analysis. The integrated safety production detection platform is responsible for the transmission and interaction of multi-source data and generates decision signals for equipment control commands based on the risk types and risk levels. It drives the work environment detection and personnel management system to dynamically adjust the safety protection area and sends active protection commands to the wearable intelligent safety detection equipment. Through this intelligent safety protection system for shipbuilding, risks can be intelligently predicted, protection can be dynamically adjusted, and shipbuilding safety can be improved.

Description

Technical Field

[0001] This invention relates to the field of shipbuilding, and in particular to an intelligent safety protection system for shipbuilding. Background Technology

[0002] Shipbuilding is characterized by high risk, long cycles, a wide range of processes, and labor intensity, making it highly susceptible to safety accidents. During the construction of large ships, accidents involving falls from heights due to a lack of safety awareness among construction workers and inadequate protective facilities rank first among all types of accidents annually. Welding, grinding, and painting operations in shipyards involve flammable and explosive gases such as oxygen, liquefied petroleum gas, and propane, leading to numerous incidents of combustion and explosions of large quantities of materials and combustibles. Furthermore, the construction process involves working in confined spaces such as liquid tanks, cargo holds, and pump rooms. Due to restricted access and lack of natural ventilation, toxic and harmful gases accumulate, or oxygen levels are insufficient, resulting in frequent incidents of poisoning and asphyxiation among personnel.

[0003] Therefore, the intelligent development of safety protection equipment, including online data collection of personnel and environment in multiple scenarios, real-time identification, location, and early warning of hazards, as well as comprehensive management and analysis of safety data, is the only way to improve the ability to predict and prevent safety in shipbuilding in the future. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes an intelligent safety protection system for ship construction, which can intelligently predict risks, dynamically adjust protection measures, and improve ship construction safety.

[0005] This invention provides an intelligent safety protection system for shipbuilding, characterized by comprising: wearable intelligent safety detection equipment, a work environment detection and personnel management system, an AI perception prediction and evaluation module, and an integrated safety production detection platform; wherein: multi-source data collected by the wearable intelligent safety detection equipment and the work environment detection and personnel management system are input into the AI ​​perception prediction and evaluation module; the AI ​​perception prediction and evaluation module generates risk types and risk levels through cross-modal analysis; the integrated safety production detection platform is responsible for the transmission and interaction of the multi-source data, and generates decision signals for equipment control commands based on the risk types and risk levels; drives the work environment detection and personnel management system to dynamically adjust the safety protection area; and sends active protection commands to the wearable intelligent safety detection equipment.

[0006] Furthermore, the wearable intelligent safety detection equipment includes: an intelligent safety helmet, an intelligent safety belt, and an intelligent wristband, configured to: collect at least a portion of the multi-source data through multi-source wearable sensors built into the intelligent safety helmet, the intelligent safety belt, and the intelligent wristband, wherein the multi-source data includes personnel physiological, movement posture, and environmental data; the work environment detection and personnel management system includes: an easily deployable environment and status detection terminal, which, for confined spaces on ships, open safety protection areas, and key protection areas in workshops, collects at least some of the multi-source data, such as gas concentration, temperature and humidity, vital signs, and audio and video data, through the fusion of multi-source environmental sensors; the AI ​​perception prediction and evaluation module generates the risk type and risk level based on the multi-source data using cross-modal analysis by an edge computing unit.

[0007] Furthermore, the operational environment detection and personnel management system also includes a safety positioning and electronic fence subsystem; within the confined space of the ship, UWB ultra-wideband technology is used to achieve real-time personnel positioning and intrusion alarms; in open / workshop protected areas, tiered expulsion warnings are triggered through RFID and photoelectric sensor linkage.

[0008] Furthermore, the AI ​​perception prediction and evaluation module employs deep learning neural network methods to detect and classify the multi-source data. It uses a comprehensive learning method combining offline data and difficult samples, supplemented by integrated hardware and software detection methods, to achieve real-time monitoring of personnel behavior and dangerous environments. A real-time data stream monitoring and early warning model is established, based on behavior and action recognition and dynamic prediction models for previous and subsequent frames. Multi-scale fusion neural networks are used to optimize the generalization parameters of the model, and combined with edge computing devices, to achieve accurate identification of personnel behavior and dangerous environments.

[0009] Furthermore, the AI ​​perception prediction and evaluation module also includes a high-precision perception AI model for multi-source data in complex environments and a dangerous abnormal sound source monitoring AI model, which uses single-model routing rules or dual-model collaborative triggering mechanisms to generate the risk type and the risk level.

[0010] Furthermore, the integrated safety production monitoring platform also collects multi-source data for safety production monitoring based on multi-protocol data transmission technology, and achieves stable transmission and interaction of the multi-source data in complex industrial environments based on heterogeneous communication mechanisms; it performs centralized processing of the multi-source data through an AI perception, prediction and evaluation module, and extracts abnormal monitoring data in conjunction with an alarm rule model, and generates the risk type and risk level; it matches safety scheduling rules to generate decision signals for equipment control commands.

[0011] Furthermore, after generating the decision signal for the equipment control command, the device is connected to the wearable smart protective equipment via a Bluetooth gateway to achieve comprehensive monitoring and warning of the working environment and the status of the personnel.

[0012] Furthermore, it also includes forming alarm thresholds and data models based on historical data and industry standards to generate decision signals for the equipment control commands.

[0013] Furthermore, the easily deployable environment and status monitoring terminal adopts a confined space adaptive deployment configuration, and compensates for signal attenuation due to metal equipment obstruction by using the ultrasonic echo delay model of the multi-source environmental sensors, thus solving the problem of blind spots in cabin corner monitoring.

[0014] Furthermore, the integrated safety production detection platform also dynamically draws personnel trajectory heat maps based on the positioning data of the safety positioning and electronic fence subsystem, and traces back high-risk operation paths.

[0015] The beneficial effects of this invention compared to existing technologies are as follows: The intelligent safety protection system for ship construction of this invention constructs a dynamic protection system covering the entire chain of "human-environment-equipment," achieving three major breakthroughs through the synergistic operation of four modules: Accurate risk prediction: Integrating multi-source data from wearable devices (personnel status) and environmental monitoring systems (work scenarios), and using AI cross-modal analysis to identify risk types and levels in real time, transforming traditional passive protection into proactive early warning; Dynamic protection response: The safety production platform drives the environmental management system to automatically adjust the protection area based on risk instructions and sends active protection instructions to personnel and equipment, so as to realize the dynamic optimization of protection strategies with risks. Integrated management efficiency: Breaking down information silos, a unified platform is used to complete the closed loop of data fusion, decision generation and command execution, significantly improving the efficiency of safety management in complex construction scenarios.

[0016] Ultimately, this achieves comprehensive benefits including reduced construction accident rates, improved emergency response speed, and optimized safety management costs; providing full-coverage, adaptive intelligent safety assurance for shipbuilding. Attached Figure Description

[0017] Figure 1 This is an overall framework diagram of the intelligent safety protection system for ship construction according to the present invention; Figure 2 This is a data composition diagram of the intelligent safety protection system for ship construction according to the present invention; Figure 3 This is a data flow diagram of the intelligent safety protection system for ship construction according to the present invention. Detailed Implementation

[0018] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0019] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0021] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0022] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0023] The present invention will now be described in detail with reference to the accompanying drawings: This invention discloses an intelligent safety protection system for ship construction, such as... Figure 1 As shown, four applications are implemented for three complex scenarios in the entire ship construction process: confined space, open safety protection areas, and key safety protection areas in the workshop. These applications achieve intelligent safety protection, operational safety monitoring, personnel management, and AI identification, forming two systems, seven types of equipment, two models, and one platform.

[0024] Specifically, the intelligent safety protection system for ship construction of this invention comprises two systems: a wearable intelligent safety detection equipment and a work environment detection and personnel management system. Multi-source data collected by the wearable intelligent safety detection equipment and the work environment detection and personnel management system are input into an AI perception, prediction, and evaluation module. This module includes two models: a high-precision perception AI model for multi-source data in complex environments and a monitoring AI model for dangerous and abnormal sound sources. Based on the specific types of the multi-source data, the AI ​​perception, prediction, and evaluation module generates risk types and risk levels through cross-modal analysis. Finally, an integrated safety production detection platform (one platform) generates decision signals for equipment control commands based on the risk types and risk levels; drives the work environment detection and personnel management system to dynamically adjust the safety protection area; and sends active protection commands to the wearable intelligent safety detection equipment. The integrated safety production detection platform is responsible for the transmission and interaction of the multi-source data.

[0025] In embodiments of this invention, an AI perception, prediction, and evaluation module responds to a multi-source data collaborative decision-making mechanism: wearable equipment (personnel status) and environmental system (space status) data are input into the AI ​​model, and a unified risk assessment (type + level) is generated through cross-modal analysis. The integrated safety production detection platform then drives hardware to perform dynamic protection. This improves response speed: the entire process from data acquisition to command execution is ≤500ms; reduces false alarm rate: multi-source cross-validation lowers the false alarm rate (e.g., safety helmet impact data + cabin oxygen concentration jointly determine the risk of unconsciousness); and achieves a closed-loop chain of "personnel status - space environment - AI decision-making - hardware execution," breaking the traditional predicament of fragmented security subsystems.

[0026] In embodiments of the present invention, wearable intelligent safety detection equipment is specifically embodied in seven types of equipment, including intelligent safety helmets, intelligent safety belts, and intelligent wristbands of wearable intelligent safety protection and quality inspection equipment systems, and easily deployable environment and status detection terminals of easily deployable environment and status detection subsystems. Wearable intelligent safety detection equipment collects personnel and environmental data and transmits it to the complex environment multi-source data high-precision perception AI model and the dangerous abnormal sound source monitoring AI model in the AI ​​perception prediction and evaluation module for analysis and processing. The results are then transmitted to the integrated safety production detection platform to realize data aggregation and display, safety early warning, and reverse control of equipment.

[0027] In embodiments of the present invention, the working environment detection and personnel management system further includes an easily deployable environment and status detection subsystem and a safety positioning and electronic fence subsystem; the safety positioning and electronic fence subsystem includes personnel positioning equipment for enclosed compartments, high-precision personnel positioning electronic fence equipment for open areas, and personnel approach and intrusion alarm equipment for protected areas; for the limited space of ships, open safety protection areas and key protection areas of workshops, multi-source data such as gas concentration, temperature and humidity, vital signs and audio and video data are collected by multi-source environmental sensors.

[0028] Furthermore, such as Figure 2 As shown, multi-source data can be categorized into personnel data, operational data, and environmental data. Personnel data specifically includes behavioral data, health data, and posture data; among which, health data includes heart rate, body temperature, and exercise status; environmental data includes light data, temperature and humidity data, gas composition data, and noise data, with gas composition data further including oxygen data, flammable and explosive data, and toxic gas data; operational data includes operational process data, equipment operation data, and operation execution data, with equipment operation data including equipment status and equipment parameters.

[0029] In the embodiments of the present invention, during the operation of the intelligent safety protection system for ship construction, data routing decisions follow the scenario-driven classification-collaborative analysis principle. Through a judgment mechanism, data is precisely allocated to either the high-precision perception AI model (hereinafter referred to as the perception model) for multi-source data in complex environments or the AI ​​model for monitoring dangerous and abnormal sound sources (hereinafter referred to as the sound source model). The specific logic is as follows (the following are some embodiments and do not limit other undescribed embodiments): I. Single-model routing rules Dedicated data for sound source models Judgment criteria: The raw data carries audio spectrum characteristics or equipment mechanical vibration characteristics; Data types: Noise data in environmental data (such as high-frequency sound patterns of metal cutting >10kHz) and equipment operation data in operational data (such as abnormal noise characteristics of host bearing friction). Example scenario: When an acoustic sensor in a sealed cabin captures a sustained 8-12kHz sound of metal fatigue crack propagation, the data packet is automatically labeled with an audio identifier, and the sound source model is directly connected to perform wavelet packet energy entropy analysis, outputting the conclusion of "structural damage risk".

[0030] Perception model-specific data Judgment criteria: The data has spatial correlation and is not audio-based; Data type: Personnel data: behavioral data (helmet posture angle > 45°), health data (wristband heart rate > 150 bpm), and posture data (UWB positioning coordinates). Environmental data: gas composition (methane concentration > 1% LEL), illumination data (arc intensity > 10,000 lux), temperature and humidity (temperature rise rate > 3℃ / min); Operation data: Operation execution data (events where welding masks were not worn); Example scenario: The tension sensor of the safety belt of a worker working at height returns 12.8kN, which exceeds the limit (numerical data). When combined with the motion characteristics of UWB positioning speed > 1.5m / s, a climbing fall risk assessment is triggered.

[0031] II. Dual-model collaborative triggering mechanism Judgment criteria: The presence of cross-modal risk coupling characteristics or the need for spatial verification of sound source events.

[0032] Sound source event triggers spatial perception (sound source model triggers perception model): When the sound source model identifies that the grinding operation is not performed without earmuffs (e.g., continuous exposure at 6.8kHz): the video analysis thread of the perception model is automatically activated, the nearest camera is called to focus on the direction of the sound source, and the earmuff wearing status is detected.

[0033] Output a composite decision: "violation of regulations + risk of hearing damage".

[0034] Environmental anomaly activates voiceprint verification (perception model triggers sound source model): When the perception model detects a benzene concentration of 0.9 ppm in the painting workshop (dynamic threshold 0.8 ppm), it sends a gas leak voiceprint matching request to the sound source model. The sound source model scans the "hissing" leak characteristics in the audio stream (such as 500-800Hz broadband noise) and outputs: "Toxic gas leak + diffusion risk".

[0035] The dual-model collaborative architecture demonstrates three core advantages in shipbuilding safety protection through specialized division of labor and cross-modal verification: First, the acoustic model focuses on processing audio stream features (such as abnormal equipment noise or leakage sound patterns), while the perception model integrates visual and environmental parameters (such as personnel posture and gas concentration). This division of labor significantly reduces the computational complexity of a single model. Second, when the acoustic model detects abnormal sounds (such as the sound of metal fatigue crack propagation), it automatically triggers the perception model to verify spatial states (such as personnel positions and equipment operating conditions), forming a multi-source evidence chain and improving the reliability of risk assessment. Third, the modular design facilitates the expansion of new sensor data streams (such as incorporating vibration spectra into the acoustic model and rerouting sudden temperature and humidity changes to the perception model), enhancing the system's adaptability to complex industrial scenarios. Essentially, this architecture decouples heterogeneous data types and establishes collaborative triggering logic, ensuring real-time performance while strengthening the robustness of risk identification.

[0036] During the data acquisition phase, visual and audio sensors are used to comprehensively collect visual and audio data related to personnel behavior, posture, and hazardous environments within the confined space of the ship. The sensor equipment transmits the collected hazardous environment data to the AI ​​model in real time. Upon receiving the data, the AI ​​model uses YOLO target detection and SORT multi-target tracking algorithms to accurately detect and track targets in the visual data, quickly identifying the position and posture of personnel. Simultaneously, the transform algorithm for abnormal sound detection analyzes and judges abnormal sounds in the audio data, such as abnormal sounds caused by equipment malfunctions or dangerous operations. For personnel behavior data, the AI ​​model uses specialized extraction techniques combined with personnel behavior confidence calculation methods to conduct in-depth analysis of personnel behavior, determining whether personnel have engaged in unauthorized operations and the standardization and safety of their actions. In this process, based on the established integrated hardware and software deep learning and sensor fusion model, visual, audio, and other multi-source sensor data are fused and processed. Through deep learning and inference of the model, comprehensive AI detection and identification in the confined space of a ship, providing reliable assurance for ship operation safety, is achieved.

[0037] It should be noted that the multi-source sensor of the present invention includes a multi-source wearable sensor and a multi-source environmental sensor; the multi-source wearable sensor is built into the wearable smart safety detection equipment such as the smart safety helmet, the smart safety belt, and the smart bracelet; the multi-source environmental sensor is deployed in the work environment detection and personnel management system.

[0038] In embodiments of this invention, the smart safety helmet integrates a pressure sensor, which can quickly detect and promptly report information upon accidental impact, buying valuable time for subsequent medical treatment. The smart safety belt not only ensures physical fixation for workers at heights but also monitors their stress state in real time. Through a built-in pressure sensor, it accurately senses the tension on the safety belt; if the tension exceeds a safety threshold, it immediately issues a warning, alerting workers and management to potential dangers. Furthermore, it can combine motion sensor data to determine if the worker is in an unstable working posture and provide timely corrective prompts. The health monitoring bracelet focuses on monitoring the worker's physical health. It continuously and in real-time collects key physiological parameters such as heart rate, blood pressure, and body temperature. When it detects an abnormally high heart rate or blood pressure fluctuations outside the normal range, it quickly issues an alarm to facilitate timely medical intervention. Simultaneously, it can record data such as the worker's steps and activity intensity, providing a basis for assessing the worker's fatigue level. Based on multi-source data such as environmental parameters, personnel movement and physiological parameters collected by wearable equipment, the data quality is improved through multi-sensor data fusion algorithms. Data mining methods are used to analyze the data to predict dangerous areas, behaviors and personnel health and fatigue status. An anomaly detection model based on machine learning and deep learning is constructed to identify various anomalies and provide support for safety decision-making.

[0039] Wearable alarm and abnormal data interaction technology focuses on identifying improper wear and abnormal operations. It utilizes image recognition and sensor data analysis to monitor equipment wearing status and identifies abnormal operations in conjunction with operational specifications. Diverse alarm methods are employed to alert workers; an efficient abnormal data interaction mechanism is established to push abnormal data in real time to worker terminals and management monitoring platforms. The platforms trigger emergency responses according to rules to ensure the safety of shipbuilding personnel.

[0040] In embodiments of the present invention, the work environment detection and personnel management system can solve safety problems in multiple scenario safety protection zones, such as difficulty in identifying personnel status, difficulty in tracking movement, difficulty in area isolation, and difficulty in identifying hazards and monitoring the environment. The easily deployable environment and status monitoring subsystem integrates multiple sensors, including gas, temperature, and audio / video sensors, enabling rapid and convenient deployment at work sites. This subsystem can collect multi-dimensional environmental data in real time and, using advanced data analysis technology, accurately identify hazards in the work environment, such as harmful gas leaks and abnormal temperatures. Simultaneously, it leverages multi-sensor data fusion algorithms to effectively improve data accuracy, providing a reliable basis for environmental monitoring. The personnel-related data it collects, after analysis of multi-scenario personnel management and protection configuration design, helps solve the challenges of personnel status identification and movement statistics, achieving real-time monitoring of personnel status and accurate recording of movement. The safety positioning and electronic fence subsystem, based on UWB ultra-wideband wireless communication technology and Bluetooth and RFID wireless carrier communication technology, develops technologies for personnel positioning in confined spaces, personnel intrusion in open areas, and personnel proximity warnings in key laboratories. It conducts research on personnel positioning and intrusion warning models, forming a series of virtual electronic fence systems. This achieves effective personnel positioning and area control based on electronic fence and UWB wireless carrier technology research.

[0041] Furthermore, within the confined space of the ship, UWB ultra-wideband technology is used to achieve real-time personnel positioning and intrusion alarms; in open / workshop protected areas, RFID and photoelectric sensors are linked to trigger graded expulsion warnings.

[0042] Through the embodiments of this invention, the differentiated positioning and protection strategy forms three core advantages tailored to the characteristics of shipbuilding scenarios: First, in confined spaces with complex electromagnetic environments, UWB technology, with its nanosecond-level pulse signals, penetrates metal structures and resists multipath interference, achieving high-precision positioning and effectively solving the positioning failure problem caused by signal attenuation in enclosed compartments using traditional Bluetooth / RFID. Second, in open areas, RFID batch identification and photoelectric sensing linkage are used, and a physical-virtual dual barrier is constructed through tag scanning combined with laser boundary projection, balancing the economy of large-scale monitoring with real-time intrusion response. Finally, a graded response mechanism (such as wristband vibration → ranging virtual boundary → acoustic alarm) avoids excessive interference with normal operations, significantly improving the operability of protective measures. Essentially, through technology adaptation and layered response design, it simultaneously overcomes two major industry challenges in shipbuilding: the reliability of positioning in confined spaces and the cost-effectiveness of protection in open areas.

[0043] In embodiments of the present invention, the AI ​​perception prediction and evaluation module employs a deep learning neural network method to detect and classify the multi-source data. It uses an offline data + difficult sample comprehensive learning method, supplemented by an integrated hardware and software detection method, to achieve real-time monitoring of personnel behavior and dangerous environments. A real-time data stream monitoring and early warning model is established. Based on behavior and action recognition and dynamic prediction models of previous and next frames, the model's generalization parameters are optimized using a multi-scale fusion neural network. Combined with edge computing devices, the accurate identification of personnel behavior and dangerous environments is achieved.

[0044] The AI ​​perception, prediction, and evaluation module achieves precise monitoring of high-risk ship scenarios through multi-level technology fusion: First, it employs offline basic training and incremental learning of difficult samples (such as adversarial generation of false detection samples of mask detachment under strong light), combined with a multi-scale spatiotemporal fusion network to simultaneously process wide-angle global perspective and helmet first-person view. It integrates features of different resolutions through spatial pyramid pooling and uses temporal convolution to analyze the continuity of climbing actions. Second, relying on a hardware-software collaborative architecture, a lightweight YOLO model is deployed at the edge of the helmet to realize mask wearing detection, regional base stations run 3D skeleton algorithms to calculate the center of gravity offset, and cloud-based collaborative voiceprint analysis completes multimodal decision-making. While ensuring the accuracy of behavior recognition, it reduces edge power consumption, significantly overcoming the industry bottlenecks of detection latency and generalization attenuation in complex ship environments.

[0045] like Figure 3The diagram shown is a data flow chart of the intelligent safety protection system for ship construction according to an embodiment of the present invention. At the ship construction site, multi-source data is collected and transmitted through wearable intelligent safety detection equipment and a work environment detection and personnel management system. The AI ​​perception, prediction, and evaluation module processes and analyzes the data, generating risk types and risk levels. An integrated safety production detection platform performs early warning judgments. For risk-free data, detection continues; for risky data, an early warning mechanism is triggered, and an emergency response is initiated. Data is stored on the integrated safety production detection platform, forming a safety detection database. Subsequent analysis, display, decision-making, and support can be performed based on data from the current or historical time periods. Finally, safety production measures are optimized based on data applications (such as data analysis and decision-making).

[0046] In the embodiments of this application, the integrated safety production monitoring platform can realize the data fusion and transmission of various monitoring devices (environmental monitoring terminals, wearable protective equipment, etc.); in terms of technical implementation, it realizes stable transmission and interaction of multi-source data in complex industrial environments based on heterogeneous communication mechanisms; it performs centralized processing of multi-source data through AI perception, prediction and evaluation modules, and extracts abnormal monitoring data in conjunction with alarm rule models, and generates decision signals containing risk types and risk levels; it matches safety scheduling rules to generate decision signals for the equipment control commands. The heterogeneous communication mechanism (heterogeneous platform) data interaction protocol aims to achieve smooth data interaction between various monitoring equipment (such as multi-sensor IoT-based detection devices, wearable smart protective equipment, etc.) and an integrated safety production monitoring platform. Data collected by these devices, including personnel behavior, posture, and hazardous environment data, is accurately transmitted to the platform through the developed heterogeneous communication mechanism data interaction protocol. By researching and developing alarm thresholds and data models, the platform analyzes and judges the data transmitted to it. When monitoring data reaches a pre-set alarm threshold, the platform responds rapidly based on the data model. Simultaneously, upon triggering an alarm, the early warning and emergency response technology platform can quickly initiate pre-alarm procedures according to a predetermined technical route. The platform establishes an early warning mechanism and provides precise guidance for emergency response. It also establishes a comprehensive data collection and analysis architecture for personnel, environment, and operations, integrating data related to personnel, environment, and operations from various equipment. A safety monitoring database system is built to store and manage various monitoring data, providing data support for subsequent analysis. Based on this, multi-dimensional safety data analysis, display, and alarm response functions are formed. This enables the platform to not only meet the needs of routine safety management and control and monitor the safety status during ship construction in real time, but also provide strong support for emergency response operations when safety accidents occur. Ultimately, it achieves close collaboration between the integrated safety production monitoring platform and various monitoring equipment, comprehensively improving the level of safety production assurance during ship construction.

[0047] In the context of intelligent safety protection systems for shipbuilding, heterogeneous communication mechanisms refer to multi-protocol converged communication systems designed to adapt to complex industrial environments. Their core lies in solving transmission challenges that cannot be addressed by a single communication technology.

[0048] The essential difference from ordinary multimode communication.

[0049] Ordinary multimode communication: simple protocol stacking, requiring manual switching of each protocol to work independently without coordination; The system employs the following heterogeneous communication mechanisms: a dynamic scheduling protocol for the intelligent decision-making center (e.g., selecting transmission paths based on data priority), data mutual verification between protocols (e.g., UWB positioning and Bluetooth RSSI jointly resist spoofing attacks), and energy collaborative management (the wristband's low-power Bluetooth wake-up 5G module only transmits during high-risk events).

[0050] In embodiments of the present invention, the integrated platform for safety production detection can also form alarm thresholds and data models based on historical data and industry standards, and generate decision signals for equipment control commands.

[0051] In embodiments of the present invention, the integrated safety production monitoring platform achieves precise risk management in shipbuilding scenarios by constructing a dynamic alarm threshold system and a data-driven decision-making model: the platform integrates historical accident databases (such as environmental parameters and equipment status records of 200 high-altitude fall incidents in a shipyard over the past five years) and industry safety standards, uses time series analysis to mine accident precursor features (such as the displacement acceleration pattern in the 10 seconds before a sudden increase in seat belt tension), and establishes a multivariate coupled risk probability model based on a Bayesian network; when real-time monitoring data streams are input, the system first dynamically adjusts the thresholds according to environmental conditions (e.g., in a closed-cell hypoxia warning, the summer threshold is set to 19.5% oxygen concentration, and the winter threshold is adjusted to 2%). (0.2% to compensate for ventilation differences), and then match the preset decision rule tree (for example, when the welding mask is detected to be detached and the arc light intensity exceeds the standard, a three-level response is automatically triggered: wristband vibration warning → activation of safety helmet sunshade → cutting off welding machine power); this mechanism significantly improves the effectiveness of alarms and avoids false alarms caused by fixed thresholds (such as a 57% reduction in the false judgment rate of conventional temperature rise caused by high temperature weather) and missed alarms (such as an 82% increase in the detection rate of early vibration characteristics of fatigue fracture of hoisting slings). At the same time, through the direct connection between decision signals and hardware control systems (such as the electronic fence radius scaling in real time according to the risk level), a closed-loop management is formed from risk perception to physical intervention, ultimately achieving the forward shift of accident prevention while ensuring the continuity of operations.

[0052] In an embodiment of the present invention, the easily deployable environment and status monitoring terminal also adopts a narrow space adaptive deployment configuration, and compensates for signal attenuation due to metal equipment obstruction by using an ultrasonic echo delay model of multi-source environmental sensors, thereby solving the problem of blind spots in cabin corner monitoring.

[0053] The easily deployable environmental and condition monitoring terminal overcomes the challenge of monitoring confined spaces on ships through innovative configuration design: targeting traditional blind spots such as cabin corners and curved compartments, the terminal adopts a modular magnetic universal bracket structure equipped with deformable articulated arms (6 degrees of freedom adjustment), allowing the sensor array to fit closely to the hull structure with a curvature radius of 0.6 meters; based on a multi-source ultrasonic sensor network, a special echo attenuation model for metallic environments is established, and the signal strength is dynamically calibrated through a time delay compensation algorithm, effectively offsetting signal distortion caused by metallic obstacles such as pipelines and beams; this solution significantly eliminates monitoring blind spots, providing blind-spot-free safety monitoring for high-risk areas of ship construction.

[0054] In the embodiments of this application, the integrated safety production monitoring platform also dynamically generates personnel trajectory heatmaps based on the positioning data of the safety positioning and electronic fence subsystem, and traces back high-risk operation paths. The integrated safety production monitoring platform achieves refined management and control of high-risk operations in shipbuilding through personnel trajectory heatmap technology: Based on the UWB high-precision location data stream provided by the safety positioning subsystem, the platform uses a kernel density estimation algorithm to dynamically generate heatmaps, visually presenting hotspot areas of personnel activity with color gradients (e.g., red indicates high-frequency paths, blue indicates low-activity areas); for confined space operation scenarios, the system automatically associates risk level areas divided by electronic fences (e.g., flammable areas in painting workshops, edge areas of high-altitude work platforms), and when continuous trajectories are detected crossing red high-risk zones (e.g., a worker approaches an unprotected porthole 7 times within 2 hours), the backtracking analysis module is automatically triggered to reconstruct the operation process through timeline playback and identify violation operation patterns (e.g., crossing restricted areas without following the prescribed path); managers can use this to optimize operating procedures (e.g., adjust passage settings, add guardrails), and implement targeted safety training for individuals. This technology transforms discrete positioning data into insights into spatial risk distribution, significantly enhancing the ability to trace potential hazards, providing data support for preventative safety management, and effectively reducing the risk of accidents caused by unreasonable route planning or violations of regulations.

[0055] This invention addresses the challenges of high safety requirements for personnel operations, difficulty in hazard identification, and challenges in multi-information fusion for safety early warning and control in complex working conditions such as confined spaces, open spaces, and key areas of workshops during shipbuilding. It conducts research on wearable intelligent safety detection equipment for multi-scenario applications, operational environment detection and personnel management systems, and an integrated platform for AI perception, prediction, evaluation, and safety production monitoring based on multi-sensor IoT. Breakthroughs are achieved in key technologies such as multi-sensor heterogeneous data interaction and fusion with edge computing, YOLO / SORT-based multi-target detection and tracking algorithms, UWB-based ultra-wideband wireless carrier communication positioning, and the construction and comprehensive application of safety information databases. Engineering prototypes are developed, including wearable intelligent protective equipment for workers (including safety helmets, safety belts, and smart bracelets), easily deployable environmental and status monitoring terminals (combining multi-sensor information collection, personnel positioning base stations, and edge computing functions), and dynamic electronic fences. A complete integrated safety detection and production platform software system is also developed. This system enables online data collection of personnel and environment in multiple scenarios, real-time identification, location, and early warning of hazard sources, and comprehensive management and analysis of safety data. Effectively enhance the ability to predict and prevent safety risks in shipbuilding, and provide equipment and system support for the intelligent development of safe ship production.

[0056] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platform, or the corresponding software can be implemented by hardware platform.

[0057] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application. Those skilled in the art will understand that the modules in the apparatus of the embodiment can be distributed within the apparatus of the embodiment as described, or can be modified to be located in one or more apparatuses different from this embodiment. The modules of the above-described embodiment can be combined into one module, or further divided into multiple sub-modules.

[0058] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of any particular implementation scenario. The above disclosures are merely a few specific implementation scenarios of this application; however, this application is not limited thereto, and any variations conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. A smart safety protection system for ship construction, characterized in that, include: Wearable intelligent safety inspection equipment, work environment inspection and personnel management system, AI perception prediction and evaluation module, and integrated safety production inspection platform; among which: The wearable intelligent safety detection equipment and the multi-source data collected by the working environment detection and personnel management system are input into the AI ​​perception prediction and evaluation module; The AI ​​perception, prediction, and evaluation module generates dynamic risk types and risk levels for the confined space and metal structure environment of ship construction through cross-modal analysis. The integrated safety production detection platform is responsible for the transmission and interaction of the multi-source data, and generates decision signals for equipment control commands based on the risk type and risk level; drives the work environment detection and personnel management system to physically and dynamically adjust the safety protection area; and sends active protection commands to the wearable intelligent safety detection equipment to trigger it to actively perform protective actions. The AI ​​perception prediction and evaluation module also includes a high-precision perception AI model for multi-source data in complex environments and a dangerous and abnormal sound source monitoring AI model. It uses a single-model routing rule based on data feature classification or a dual-model collaborative triggering mechanism that includes a bidirectional active triggering verification relationship between the sound source model and the perception model to generate the risk type and the risk level. The single-model routing rule includes automatically selecting the processing model according to the type of input data, and the dual-model collaborative triggering mechanism includes establishing a bidirectional triggering verification relationship between the sound source model and the perception model in a specific risk scenario.

2. The intelligent safety protection system for ship construction according to claim 1, characterized in that: The wearable smart safety detection equipment includes: a smart safety helmet, a smart safety belt, and a smart bracelet, configured as follows: At least a portion of the multi-source data is collected by the multi-source wearable sensors built into the smart helmet, the smart seat belt, and the smart bracelet, wherein the multi-source data includes personnel physiological, movement posture, and environmental data; The operational environment detection and personnel management system includes: an easily deployable environment and status detection terminal, which collects multi-source data, including gas concentration, temperature and humidity, vital signs, and audio-visual data, as well as other data, for confined spaces on ships, open safety protection areas, and key protection areas in workshops through the fusion of multi-source environmental sensors; and an AI perception, prediction, and evaluation module that generates the risk type and risk level based on the multi-source data using cross-modal analysis by an edge computing unit.

3. The intelligent safety protection system for ship construction according to claim 2, characterized in that: The work environment detection and personnel management system also includes a safety positioning and electronic fence subsystem; Within the confined space of a ship, UWB (Ultra-Wideband) technology is used to achieve real-time personnel positioning and intrusion alarms; In open / workshop protected areas, tiered expulsion warnings are triggered by the linkage of RFID and photoelectric sensors.

4. The intelligent safety protection system for ship construction according to claim 2, characterized in that: The AI ​​perception, prediction, and evaluation module employs deep learning neural network methods to detect and classify the multi-source data. It uses a comprehensive learning method combining offline data and difficult samples, supplemented by integrated hardware and software detection techniques, to achieve real-time monitoring of personnel behavior and hazardous environments. A real-time data stream monitoring and early warning model is established, based on behavior and action recognition and dynamic prediction models for previous and subsequent frames. Multi-scale fusion neural networks are used to optimize the model's generalization parameters, and edge computing devices are combined to achieve accurate identification of personnel behavior and hazardous environments.

5. The intelligent safety protection system for ship construction according to claim 1, characterized in that: The integrated platform for safety production monitoring also collects multi-source data for safety production monitoring based on multi-protocol data transmission technology, and achieves stable transmission and interaction of the multi-source data in complex industrial environments based on heterogeneous communication mechanisms. The multi-source data is centralized through the AI ​​perception, prediction, and evaluation module. Combined with the alarm rule model, abnormal monitoring data is extracted, and the risk type and risk level are generated. The decision signal for the equipment control command is generated by matching the safety scheduling rules.

6. The intelligent safety protection system for ship construction according to claim 5, characterized in that: After generating the decision signal for the equipment control command, the device is connected to wearable smart protective equipment via a Bluetooth gateway to achieve comprehensive monitoring and alerts of the working environment and the status of the personnel.

7. The intelligent safety protection system for ship construction according to claim 5, characterized in that: It also includes forming alarm thresholds and data models based on historical data and industry standards, and generating decision signals for the equipment control commands.

8. The intelligent safety protection system for ship construction according to claim 2, characterized in that: The easily deployable environment and status monitoring terminal adopts a confined space adaptive deployment configuration. It compensates for signal attenuation caused by metal equipment obstruction through the ultrasonic echo delay model of the multi-source environmental sensors, thus solving the problem of blind spots in the monitoring of cabin corners.

9. The intelligent safety protection system for ship construction according to claim 3, characterized in that: The integrated safety production detection platform also dynamically draws personnel trajectory heat maps based on the positioning data of the safety positioning and electronic fence subsystem, and traces back high-risk operation paths.

Images