Special operation intelligent monitoring and safety risk assessment system and method based on visual AI

Abstract

The invention discloses a special operation intelligent monitoring and safety risk assessment system and method based on visual AI, and is applied to the technical field of special operation safety monitoring. The system comprises a data acquisition module for acquiring the concentration of poisonous and harmful gas and monitoring video data; the data transmission module is used for transmitting data; the alarm identification module is used for identifying danger and giving an alarm based on the real-time poisonous and harmful gas concentration and monitoring video data; the operation alarm module is used for carrying out sound-light alarm reminding on equipment and displaying alarm information through a software platform; the operation risk assessment module is used for performing operation risk assessment based on multi-dimensional operation related risk factors; and the operation monitoring module is used for monitoring the operation process. The safety monitoring efficiency and the accident prevention capability of special operation are comprehensively improved by integrating real-time toxic and harmful gas monitoring, artificial intelligence visual analysis and the 5G wireless communication technology.

Description

Technical Field

[0001] The present invention relates to the technical field of special operation safety monitoring technology, and more specifically to a special operation intelligent monitoring and safety risk assessment system and method based on visual AI. Background Art

[0002] In special working environments, such as work sites in the chemical, mining, and electric power industries, there are many safety risks. Traditional monitoring methods often rely on manual inspections, which are inefficient and prone to omissions. Some existing safety monitoring equipment has relatively simple functions and mainly relies on manual monitoring or a single sensor for simple threshold alarms. It has the defects of a single monitoring dimension (only basic parameters such as gas and temperature), and is unable to identify complex personnel behaviors (such as not wearing safety equipment, illegal operations) and environmental anomalies (such as flames, missing tools). It is difficult to meet the needs of comprehensive monitoring of personnel behavior and environmental conditions in complex and special working environments. It is also impossible to collaboratively and dynamically evaluate operational safety risks from multiple dimensions. It is difficult to detect and deal with situations such as toxic and harmful gas leaks, illegal operations by personnel, and abnormal environmental changes in a timely manner, resulting in frequent safety accidents and causing huge losses to human life and corporate property. Therefore, how to provide a special operation intelligent monitoring and safety risk assessment system and method based on visual AI is a problem that technicians in this field urgently need to solve. Summary of the Invention

[0003] In view of this, the present invention provides a special operation intelligent monitoring and safety risk assessment system and method based on visual AI, which can conduct real-time comprehensive monitoring and dynamic risk assessment of toxic and harmful gas leaks, personnel behavior and environmental anomalies in high-risk special operation scenarios such as chemical, mining, and electricity, and realize advance warning and active prevention and control of safety risks.

[0004] In order to achieve the above object, the present invention provides the following technical solutions:

[0005] A special operation intelligent monitoring and safety risk assessment system based on visual AI, including a data acquisition module, a data transmission module, an alarm identification module, an operation alarm module, an operation risk assessment module and an operation monitoring module; the data acquisition module collects toxic and harmful gas concentrations and monitoring video data, the data transmission module is used to transmit data, the alarm identification module identifies dangers and alarms based on real-time toxic and harmful gas concentrations and monitoring video data, and pushes alarm information to the operation alarm module. The operation alarm module issues equipment sound and light alarm reminders and displays alarm information through the software platform at the same time. The operation risk assessment module performs operation risk assessment based on multi-dimensional operation-related risk factors, and the operation monitoring module is used to monitor the operation process.

[0006] Optionally, the data acquisition module includes a toxic and harmful gas acquisition unit, a monitoring video acquisition unit and a battery charging management unit; the toxic and harmful gas acquisition unit extracts gas from the on-site environment and transports it to the toxic and harmful gas sensor for real-time sample collection, the monitoring video acquisition unit continuously collects on-site visible light and infrared video image data at a set frame rate at the work site, and the battery charging management unit includes the main device battery and the video camera battery to provide power for gas pump acquisition and video monitoring.

[0007] Optionally, the data transmission module uses 5G wireless communication, and the gas concentration time series data and visible light and thermal infrared monitoring videos collected at the work site are transmitted to the software platform in real time through the 5G network.

[0008] Optionally, the alarm identification module includes a gas anomaly identification unit, an on-site operation risk identification unit and an alarm information push unit; the gas anomaly identification unit reads the real-time toxic and harmful gas data collected by the data acquisition module, and determines whether to alarm based on the set fixed threshold. When the real-time monitoring value ≥ the fixed threshold, it is determined to be a gas anomaly. The on-site operation risk identification unit identifies the anomaly in the image based on the pre-trained YOLOv8 model. If the duration of the anomaly is greater than the set threshold time, it is determined to be an alarm. The alarm information push unit pushes the generated real-time alarm information to the operation alarm module via WebSocket in real time. The pushed information includes the alarm type, alarm time, alarm level, alarm value and alarm image.

[0009] Optional abnormal situations include flames, people falling to the ground, not wearing safety helmets, not wearing protective clothing, not wearing reflective clothing, playing with mobile phones, missing fire extinguishers, not using double hooks when working at heights, throwing objects from heights when working at heights, not wearing personal protective equipment when working with temporary electricity, not setting up circuit breaker signs when working with circuit breakers, and not setting up blind plate signs when working with blind plates.

[0010] Optionally, the operation alarm module includes an equipment sound and light alarm unit and an alarm information management unit; the equipment sound and light alarm unit receives the alarm information pushed by the alarm identification module, displays the abnormality type and related data through the digital display screen of the portable integrated monitoring device, and issues a voice broadcast at the same time to inform the operating personnel and management personnel of the specific abnormal situation, triggering the sound and light alarm, attracting the attention of on-site personnel through flashing lights and loud alarm sounds, and reminding them to take corresponding safety measures. The alarm information management unit receives the alarm information pushed by the alarm identification module, and stores and manages the alarm information.

[0011] Optionally, the operation risk assessment module includes an indicator determination unit, an indicator value calculation unit, a weight determination unit, a risk value calculation unit, and an operation risk grading unit; the indicator determination unit considers the operation risk and selects the operation type, the number of operation alarms, the operation environment, the cross operation, and the number of people in the operation area from three dimensions: operation type, operation alarm, and environmental risk as risk indicators; the indicator value calculation unit determines its indicator value based on the risk indicator data; the weight determination unit determines the weight value of each risk indicator; and the risk value calculation unit calculates the risk value S:

[0012] S=s1w1+s2w2+s3w3+s4w4+s5w5

[0013] Where s1, s2, s3, s4, and s5 are the index values ​​of operation type, number of operation alarms, operation environment, cross-operation, and number of people in the operation area, respectively; w1, w2, w3, w4, and w5 are the weight values ​​of s1, s2, s3, s4, and s5, respectively; the operation risk grading unit divides the operation into different risk levels based on the risk value S.

[0014] Optionally, the operation monitoring module comprehensively displays a variety of real-time monitoring information, providing full-process support for managers to conduct remote operation monitoring. Specific information displayed includes real-time monitoring video collected during the operation, current real-time gas concentration, historical gas concentration changes, historical alarm records and current risk level.

[0015] A method for intelligent monitoring and safety risk assessment of special operations based on visual AI is applied to the above-mentioned intelligent monitoring and safety risk assessment system for special operations based on visual AI, comprising the following steps:

[0016] The data acquisition module collects toxic and harmful gas concentrations and monitoring video data;

[0017] The data transmission module transmits the data to the alarm identification module;

[0018] The alarm recognition module identifies dangers and issues alarms based on real-time toxic and hazardous gas concentrations and surveillance video data, and pushes the alarm information to the operation alarm module;

[0019] The operation alarm module provides equipment sound and light alarm reminders and displays alarm information through the software platform;

[0020] The operation risk assessment module conducts operation risk assessment based on multi-dimensional operation-related risk factors;

[0021] The job monitoring module displays real-time monitoring information.

[0022] Through the above technical solutions, it can be seen that compared with the existing technology, the present invention provides a special operation intelligent monitoring and safety risk assessment system and method based on visual AI, which has the following beneficial effects:

[0023] 1. Identify and warn of operational risks based on visual artificial intelligence: During operations, the system can automatically identify unsafe behaviors, abnormal movements, and potential safety hazards based on surveillance video. Combined with technologies such as the Internet of Things, it can provide all-round, multi-dimensional intelligent monitoring of special work sites, greatly improving the comprehensiveness and accuracy of safety monitoring. It can identify and warn of 18 types of operational risks by combining the characteristics of eight special operations and the actual needs of the work scenarios, with an identification accuracy rate exceeding 90%. Compared with traditional methods, this system improves safety supervision efficiency and assists relevant personnel in real-time monitoring of on-site operation dynamics and alarm status, enabling rapid coordinated response and significantly improving accident handling efficiency.

[0024] 2. Dynamic operation safety risk assessment based on multi-factor integration: This invention integrates static operation attribute information and dynamic risk factors to construct a multi-dimensional dynamic safety risk assessment model, effectively assessing the degree of dynamic safety risk of operations, and providing strong support for dynamic supervision of operation safety risks and the implementation of safety control measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0026] Figure 1 This is a schematic diagram of the special operation intelligent monitoring and safety risk assessment system of the present invention. DETAILED DESCRIPTION

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0028] The embodiment of the present invention discloses a special operation intelligent monitoring and safety risk assessment system based on visual AI, such as Figure 1As shown, it includes a data acquisition module, a data transmission module, an alarm identification module, an operation alarm module, an operation risk assessment module and an operation monitoring module; the data acquisition module collects toxic and harmful gas concentrations and monitoring video data, the data transmission module is used to transmit data, the alarm identification module identifies dangers and alarms based on real-time toxic and harmful gas concentrations and monitoring video data, and pushes alarm information to the operation alarm module. The operation alarm module performs equipment sound and light alarm reminders and displays alarm information through the software platform. The operation risk assessment module performs operation risk assessment based on multi-dimensional operation-related risk factors, and the operation monitoring module is used to monitor the operation process.

[0029] Furthermore, the data acquisition module includes a toxic and harmful gas acquisition unit, a monitoring video acquisition unit and a battery charging management unit; the toxic and harmful gas acquisition unit extracts gas from the on-site environment and transports it to the toxic and harmful gas sensor for real-time sample collection, and the monitoring video acquisition unit continuously collects on-site video image data at a set frame rate at the work site, and the battery charging management unit: the battery charging management unit includes a main device battery and a video camera battery, which provide power for gas pump acquisition and video monitoring.

[0030] In the embodiment of the present invention, the toxic and harmful gas sensor can accurately detect H2S, CO, O2, NH3, NO x The monitoring video acquisition unit adopts an explosion-proof camera, which provides two convenient installation methods: bracket fixation and strong magnetic base, which allows operators to flexibly adjust the camera position according to actual working conditions; the battery adopts a large-capacity rechargeable battery of 12V / 30AH.

[0031] Furthermore, the data transmission module uses 5G wireless communication, and the gas concentration time series data and visible light and thermal infrared monitoring videos collected at the work site are transmitted to the software platform in real time through the 5G network.

[0032] Furthermore, the alarm identification module includes a gas anomaly identification unit, an on-site operation risk identification unit and an alarm information push unit; the gas anomaly identification unit reads the real-time toxic and harmful gas data collected by the data acquisition module, and determines whether to alarm according to the set fixed threshold. When the real-time monitoring value ≥ the fixed threshold, it is determined to be a gas anomaly. The on-site operation risk identification unit identifies abnormal conditions in the image frames of the video based on the pre-trained YOLOv8 model. If the duration of the abnormal condition is greater than the set threshold time, it is determined to be an alarm. The alarm information push unit pushes the generated real-time alarm information to the operation alarm module through WebSocket in real time. The pushed information includes alarm type, alarm time, alarm level, alarm value and alarm picture.

[0033] Furthermore, abnormal situations include flames, people falling to the ground, not wearing safety helmets, not wearing protective clothing, not wearing reflective clothing, playing with mobile phones, missing fire extinguishers, not using double hooks when working at heights, throwing objects from heights when working at heights, not wearing personal protective equipment when working with temporary electricity, not setting up circuit breaker signs when working with circuit breakers, and not setting up blind plate signs when working with blind plates.

[0034] In the embodiment of the present invention, the gas types identified by the gas anomaly identification unit include H2S, CO, O2, NH3, NO x (nitrogen oxides) and EX (combustible gases), the alarm thresholds of various gases are determined in accordance with national and industry standards;

[0035] The training process of the YOLOv8 model is:

[0036] Feature data collection: We collect images of normal and abnormal operations in various special operation scenarios. Abnormal operation images include flames, people falling to the ground, workers not wearing safety helmets, protective clothing, or reflective clothing, people using mobile phones, missing fire extinguishers, workers not using double hooks when working at heights, objects thrown from heights during work at heights, people not wearing personal protective equipment during temporary power operations, circuit breaker operations without circuit breaker signs, and blind plate operations without blind plate signs. Operation images are obtained through open source datasets and on-site collection.

[0037] Data annotation: Use data annotation tools to annotate the collected images, clearly marking the abnormality type and location;

[0038] Image preprocessing: Based on the annotation information, we filter out images containing specific anomalies. In addition, for dimly lit and blurry images, we use methods such as histogram equalization and contrast stretching to enhance image clarity and contrast, highlighting the target object. We also use algorithms such as Gaussian filtering and median filtering to remove noise interference in the image and improve image quality.

[0039] Model training: Divide the labeled image dataset into training, validation, and test sets in an 8:1:1 ratio. Use the YOLOv8 model to train the model based on the training set, and continuously adjust the model parameters through backpropagation. Use the validation set to evaluate model performance and adjust the training strategy in a timely manner to prevent overfitting. Use the test set to evaluate the final model performance to ensure that the model has good generalization ability and robustness.

[0040] The risk event identification process of the YOLOv8 model is as follows:

[0041] Real-time video stream processing: Using video stream processing technology, the real-time video stream collected by the data acquisition module is decoded and frame extracted. At the same time, the extracted video frames are pre-processed, including image resizing and normalization, to meet the input requirements of the model.

[0042] Anomaly Identification: Input pre-processed video frames into the trained risk identification model to quickly identify targets such as people, equipment, and objects in the video frames, and determine the category and location of abnormal events;

[0043] Risk Alarm: Based on historical frame detection results, if the duration of an anomaly detection exceeds the set threshold, an alarm is triggered to avoid instantaneous misjudgments. For example, if a helmet is not worn for 5 seconds, a missing helmet alarm is generated. Alarm types include flames, falling, not wearing a helmet, not wearing protective clothing, not wearing reflective clothing, playing with a mobile phone, missing fire extinguisher, not using double hooks when working at height, throwing objects from a height when working at height, not wearing personal protective equipment when working with temporary power, not setting a circuit breaker sign when working with a circuit breaker, and not setting a blind plate sign when working with a blind plate.

[0044] Furthermore, the operation alarm module includes an equipment sound and light alarm unit and an alarm information management unit; the equipment sound and light alarm unit receives the alarm information pushed by the alarm identification module, displays the abnormality type and related data through the digital display screen of the portable integrated monitoring device, and issues a voice broadcast at the same time to inform the operating personnel and management personnel of the specific abnormal situation, triggering the sound and light alarm, and attracting the attention of on-site personnel through flashing lights and loud alarm sounds, reminding them to take corresponding safety measures. The alarm information management unit receives the alarm information pushed by the alarm identification module, and stores and manages the alarm information.

[0045] Furthermore, the operation risk assessment module includes an indicator determination unit, an indicator value calculation unit, a weight determination unit, a risk value calculation unit, and an operation risk grading unit; the indicator determination unit considers the operation risk and selects the operation type, the number of operation alarms, the operation environment, the cross operation, and the number of people in the operation area from the three dimensions of operation type, operation alarm, and environmental risk as risk indicators; the indicator value calculation unit determines its indicator value based on the risk indicator data; the weight determination unit determines the weight value of each risk indicator; and the risk value calculation unit calculates the risk value S:

[0046] S=s1w1+s2w2+s3w3+s4w4+s5w5

[0047] Where s1, s2, s3, s4, and s5 are the index values ​​of operation type, number of operation alarms, operation environment, cross-operation, and number of people in the operation area, respectively; w1, w2, w3, w4, and w5 are the weight values ​​of s1, s2, s3, s4, and s5, respectively; the operation risk grading unit divides the operation into different risk levels based on the risk value S.

[0048] In the embodiment of the present invention, the indicator value calculation unit calculates the indicator value specifically as follows:

[0049] Job type (s1): Determine the corresponding indicator value according to the job type. The job type indicator values ​​are shown in Table 1:

[0050] Table 1. Index values ​​of job types

[0051]

[0052] Operation alarm (s2): The operation alarm score is calculated based on the alarm type and the number of alarms. The accumulated value is less than or equal to 10 points. The number of alarms is the cumulative number of alarms from the start of the operation to the current time (if the alarm is continuous, it is only counted once). The operation alarm score is shown in Table 2:

[0053] Table 2 Operation alarm scores

[0054]

[0055] Working environment (s3): Determine the corresponding indicator value based on the working area environment, including whether it is in a major hazard source area or in a flammable and explosive area. The corresponding field values ​​are selected when reading the work ticket. The working environment indicator values ​​are shown in Table 3:

[0056] Table 3 Working environment index values

[0057]

[0058] Cross-operation (s4): Determine the corresponding index value based on the number of cross-operations. The number of cross-operations reads the cross-operation ticket selected when the job ticket is submitted and calculates the number of cross-operations. The index value of the number of cross-operations is shown in Table 4:

[0059] Table 4 Cross-operation quantity index values

[0060] Number of cross-operations Index value There are ≥2 types of special operations at the same time 5 No cross-tasking 0

[0061] Number of people in the operating area (s5): The corresponding indicator value is determined based on the number of people in the operating area. The number of people in the operating area needs to obtain the real-time coordinates of the personnel based on the personnel positioning card carried by the operator. Based on the electronic fence of the operating area drawn when the ticket is raised, the number of people in the operating area is calculated using GIS tools. The indicator value of the number of people in the operating area is shown in Table 5:

[0062] Table 5 Number of people in the working area

[0063] Number of people in the working area Index value ≥9 people 10 4≤Number of people<9 8 1≤Number of people<4 5 0 people 0

[0064] In the embodiment of the present invention, the weight values ​​of the indicators determined by the weight determination unit are shown in Table 6:

[0065] Table 6 Weight values ​​of each indicator

[0066]

[0067] In this embodiment of the present invention, the operation risk grading unit categorizes operations into different risk levels based on the current risk value S of the operation. The risk level is positively correlated with the S value; the higher the risk value, the greater the risk level. Enterprises can formulate differentiated safety supervision mechanisms based on the current risk level of the operation, including safety inspection frequency, emergency response mechanism, etc. The risk level classification is shown in Table 7:

[0068] Table 7 Risk level classification

[0069] score Risk Level Risk level 9-10 Level I Danger 6-8 Level II More dangerous 3-5 Level III Safer 0-2 Level IV Safety

[0070] Furthermore, the operation monitoring module comprehensively displays a variety of real-time monitoring information, providing full-process support for managers to conduct remote operation monitoring. The specific displayed information includes real-time monitoring videos collected during the operation, current real-time gas concentration, historical gas concentration changes, historical alarm records and current risk levels.

[0071] In an embodiment of the present invention, the display scheme of the operation monitoring module is specifically as follows:

[0072] Real-time monitoring video: displays the real-time video monitoring data collected by the data acquisition module and plays it through video stream;

[0073] Real-time gas concentration: displays real-time toxic and harmful gas data collected by the data acquisition module;

[0074] Historical gas concentration changes: Stores the real-time toxic and harmful gas data collected by the data acquisition module, and displays the gas concentration change trend from the start time of the operation to the current time through a data line chart;

[0075] Historical alarm records: read the historical alarm records stored in the job alarm module and display them;

[0076] Current Risk Level: Read the risk level assessed by the job risk assessment module.

[0077] and Figure 1 Corresponding to the system shown, an embodiment of the present invention further discloses a method for intelligent monitoring and safety risk assessment of special operations based on visual AI, which is applied to the above-mentioned intelligent monitoring and safety risk assessment system for special operations based on visual AI, including the following steps:

[0078] The data acquisition module collects toxic and harmful gas concentrations and monitoring video data;

[0079] The data transmission module transmits the data to the alarm identification module;

[0080] The alarm recognition module identifies dangers and issues alarms based on real-time toxic and hazardous gas concentrations and surveillance video data, and pushes the alarm information to the operation alarm module;

[0081] The operation alarm module provides equipment sound and light alarm reminders and displays alarm information through the software platform;

[0082] The operation risk assessment module conducts operation risk assessment based on multi-dimensional operation-related risk factors;

[0083] The job monitoring module displays real-time monitoring information.

[0084] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. The methods disclosed in the embodiments are described briefly because they correspond to the systems disclosed in the embodiments. For relevant parts, refer to the description of the systems.

[0085] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A special operation intelligent monitoring and safety risk assessment system based on visual AI, characterized by: It includes data acquisition module, data transmission module, alarm identification module, operation alarm module, operation risk assessment module and operation monitoring module; the data acquisition module collects toxic and harmful gas concentrations and monitoring video data, the data transmission module is used to transmit data, the alarm identification module identifies dangers and alarms based on real-time toxic and harmful gas concentrations and monitoring video data, and pushes alarm information to the operation alarm module. The operation alarm module issues equipment sound and light alarm reminders and displays alarm information through the software platform. The operation risk assessment module performs operation risk assessment based on multi-dimensional operation-related risk factors, and the operation monitoring module is used to monitor the operation process.

2. The special operation intelligent monitoring and safety risk assessment system based on visual AI according to claim 1 is characterized in that: The data acquisition module includes a toxic and harmful gas acquisition unit, a monitoring video acquisition unit and a battery charging management unit; the toxic and harmful gas acquisition unit extracts gas from the on-site environment and transports it to the toxic and harmful gas sensor for real-time sample collection. The monitoring video acquisition unit continuously collects on-site visible light and infrared video image data at a set frame rate at the work site. The battery charging management unit: The battery charging management unit includes the main device battery and the video camera battery, which provide power for gas pump acquisition and video monitoring.

3. The special operation intelligent monitoring and safety risk assessment system based on visual AI according to claim 1 is characterized in that: The data transmission module uses 5G wireless communication, and the gas concentration time series data and visible light and thermal infrared monitoring videos collected at the work site are transmitted to the software platform in real time through the 5G network.

4. The special operation intelligent monitoring and safety risk assessment system based on visual AI according to claim 1 is characterized in that: The alarm identification module includes a gas anomaly identification unit, a field operation risk identification unit and an alarm information push unit; the gas anomaly identification unit reads the real-time toxic and harmful gas data collected by the data acquisition module, and determines whether to alarm based on the set fixed threshold. When the real-time monitoring value ≥ the fixed threshold, it is judged as a gas anomaly. The field operation risk identification unit identifies abnormal conditions in the image frames of the video based on the pre-trained YOLOv8 model. If the duration of the abnormal condition is greater than the set threshold time, it is judged as an alarm. The alarm information push unit pushes the generated real-time alarm information to the operation alarm module through WebSocket in real time. The pushed information includes alarm type, alarm time, alarm level, alarm value and alarm picture.

5. The special operation intelligent monitoring and safety risk assessment system based on visual AI according to claim 4 is characterized in that: Abnormal risks in on-site operations include flames, people falling to the ground, not wearing safety helmets, not wearing protective clothing, not wearing reflective clothing, playing with mobile phones, missing fire extinguishers, not using double hooks when working at heights, throwing objects from high places when working at heights, not wearing personal protective equipment when working with temporary electricity, not setting up circuit breaker signs when working with circuit breakers, and not setting up blind plate signs when working with blind plates.

6. The special operation intelligent monitoring and safety risk assessment system based on visual AI according to claim 1 is characterized in that: The operation alarm module includes an equipment sound and light alarm unit and an alarm information management unit; the equipment sound and light alarm unit receives the alarm information pushed by the alarm identification module, displays the abnormality type and related data through the digital display screen of the portable integrated monitoring device, and issues a voice broadcast at the same time to inform the operating personnel and management personnel of the specific abnormal situation, triggering the sound and light alarm, and attracting the attention of on-site personnel through flashing lights and loud alarm sounds, reminding them to take corresponding safety measures. The alarm information management unit receives the alarm information pushed by the alarm identification module, and stores and manages the alarm information.

7. The special operation intelligent monitoring and safety risk assessment system based on visual AI according to claim 1 is characterized in that: The operation risk assessment module includes an indicator determination unit, an indicator value calculation unit, a weight determination unit, a risk value calculation unit, and an operation risk grading unit. The indicator determination unit considers the operation risk and selects the operation type, the number of operation alarms, the operation environment, the cross operation, and the number of people in the operation area as risk indicators from the three dimensions of operation type, operation alarm, and environmental risk. The indicator value calculation unit determines its indicator value based on the risk indicator data. The weight determination unit determines the weight value of each risk indicator. The risk value calculation unit calculates the risk value S: S=s1w1+s2w2+s3w3+s4w4+s5w5 Where s1, s2, s3, s4, and s5 are the index values ​​of operation type, number of operation alarms, operation environment, cross-operation, and number of people in the operation area, respectively; w1, w2, w3, w4, and w5 are the weight values ​​of s1, s2, s3, s4, and s5, respectively; the operation risk grading unit divides the operation into different risk levels based on the risk value S.

8. The special operation intelligent monitoring and safety risk assessment system based on visual AI according to claim 1 is characterized in that: The operation monitoring module comprehensively displays a variety of real-time monitoring information, providing full-process support for managers to conduct remote operation monitoring. Specific information displayed includes real-time monitoring videos collected during the operation, current real-time gas concentration, historical gas concentration changes, historical alarm records and current risk levels.

9. A method for intelligent monitoring and safety risk assessment of special operations based on visual AI, characterized in that: A visual AI-based intelligent monitoring and safety risk assessment system for special operations as described in any one of claims 1 to 8 comprises the following steps: The data acquisition module collects toxic and harmful gas concentrations and monitoring video data; The data transmission module transmits the data to the alarm identification module; The alarm recognition module identifies dangers and issues alarms based on real-time toxic and hazardous gas concentrations and surveillance video data, and pushes the alarm information to the operation alarm module; The operation alarm module provides equipment sound and light alarm reminders and displays alarm information through the software platform; The operation risk assessment module conducts operation risk assessment based on multi-dimensional operation-related risk factors; The job monitoring module displays real-time monitoring information.

Images