A major hazard source risk assessment method and system based on multi-modal analysis

Abstract

The application provides a major hazard source risk assessment method and system based on multi-modal analysis, which determines a major hazard source as an evaluation target, acquires multi-modal monitoring data of the evaluation target, and determines a comprehensive risk value of the evaluation target according to the multi-modal monitoring data; and then determines a risk level of the evaluation target according to the comprehensive risk value of the evaluation target, so as to quickly and accurately determine the risk level of the major hazard source, realize real-time evaluation analysis and display of the safety risk of the major hazard source, and support instant automatic sending, checking, feedback and supervision of the early warning information according to the early warning level. That is, the method provided by the application can quickly and accurately realize risk assessment of the major hazard source, improve accuracy and timeliness, and meet the risk assessment needs of users in different use scenarios.

Description

Technical Field

[0001] This invention relates to the field of intelligent industrial safety management and control technology, and in particular to a method and system for risk assessment of major hazard sources based on multimodal analysis. Background Technology

[0002] Current risk management and early warning technologies for major hazard sources are gradually shifting towards intelligent and digital approaches. Existing technologies primarily utilize IoT sensor networks to collect key parameters such as temperature, pressure, and leakage concentration in real time, combining this with big data analysis to construct risk assessment indicator systems (such as the Analytic Hierarchy Process (AHP) and fuzzy comprehensive evaluation). Some systems incorporate machine learning algorithms (such as LSTM and random forest) to mine features from historical accident data. Regarding early warning mechanisms, most solutions employ multi-level threshold-triggered alarm modes, and some advanced systems have integrated GIS visualization platforms and emergency response databases. In recent years, the fusion of multi-source heterogeneous data (such as equipment status data, meteorological and environmental data, and video surveillance data) and the optimization of dynamic risk assessment models have become research hotspots.

[0003] However, the following shortcomings still exist in the relevant technologies: Insufficient data integration and real-time performance: Traditional methods rely on single sensors or manual inspections, and lack an efficient fusion mechanism for multi-source heterogeneous data (such as equipment status, environmental parameters, and video surveillance), resulting in incomplete risk feature extraction and high data update delays, making it difficult to capture dynamic risks in a timely manner; Furthermore, the early warning models in the relevant technologies have weak generalization capabilities, poor adaptability to complex scenarios (such as multi-hazard coupling and equipment nonlinear failure), high false alarm / false alarm rates, and model training relies on historical data, making it difficult to cope with sudden abnormal operating conditions; In addition, the relevant technologies only assess a single hazard independently, without fully considering the dynamic correlation of risks from multiple dimensions such as equipment, environment, and human operation, resulting in insufficient accuracy in global risk prediction; Moreover, they lack self-learning and optimization capabilities, cannot adjust risk weights according to real-time operating conditions, and the early warning response strategies are fixed, making it difficult to adapt to long-term evolving risks such as changes in production processes or equipment aging. Therefore, there are problems with poor accuracy and timeliness in the risk assessment methods for major hazard sources.

[0004] Therefore, there is an urgent need for a risk assessment method and system for major hazard sources based on multimodal analysis, which can quickly and accurately assess the risks of major hazard sources, improve accuracy and timeliness, and meet the risk assessment needs of users in different usage scenarios. Summary of the Invention

[0005] This invention provides a method and system for risk assessment of major hazard sources based on multimodal analysis, which can quickly and accurately assess the risks of major hazard sources, improve accuracy and timeliness, and meet the risk assessment needs of users in different usage scenarios.

[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0007] In a first aspect, the present invention provides a method for risk assessment of major hazard sources based on multimodal analysis, comprising: acquiring monitoring data of the target to be assessed; the target to be assessed includes multiple hazard sources, including at least one of storage tanks, equipment, and hazardous chemical warehouses; the monitoring data includes real-time monitoring data and alarm data of the corresponding hazard source's liquid level, temperature, pressure, and flammable / toxic gas concentration; and determining, based on the monitoring data of the target to be assessed, a process production risk coefficient, a workplace flammable / toxic gas risk coefficient, a management performance and hazard mitigation correction coefficient, and a video AI and alarm analysis early warning coefficient, wherein the process production risk coefficient is used to characterize the degree of risk present in the process production of the hazard source, and the workplace... The combustible / toxic gas risk coefficient is used to characterize the risk level of combustible / toxic gases in the workplace of the hazard source. The management performance and hazard management correction coefficient is used to characterize the impact of management performance and hazard management on the risk of the hazard source. The video AI and alarm analysis early warning coefficient is used to characterize the risk level present in the video data of the hazard source. The comprehensive risk value of the target to be evaluated is determined based on the process production risk coefficient, the combustible / toxic gas risk coefficient of the workplace, the management performance and hazard management correction coefficient, and the video AI and alarm analysis early warning coefficient. The risk level of the target to be evaluated is determined based on the comprehensive risk value, where different risk levels correspond to different numerical ranges of comprehensive risk values.

[0008] Based on the above technical solution, the present invention can be further improved as follows.

[0009] Furthermore, the risk levels include major risk, significant risk, general risk, and low risk. The risk level of the target to be assessed is determined based on its comprehensive risk value, including: if the comprehensive risk value of the target is greater than or equal to the first risk value, the risk level is determined to be major risk; if the comprehensive risk value of the target is greater than or equal to the second risk value but less than the first risk value, the risk level is determined to be significant risk; if the comprehensive risk value of the target is greater than or equal to the third risk value but less than the second risk value, the risk level is determined to be general risk; and if the comprehensive risk value of the target is less than the third risk value, the risk level is determined to be low risk.

[0010] Furthermore, the formula for determining the comprehensive risk value R is:

[0011] ；

[0012] Among them, R 固有 R represents the inherent risk baseline of the target to be evaluated.工艺 R represents the risk coefficient of the target to be evaluated in the production process. 场所 γ represents the flammable / toxic gas risk factor of the workplace where the target is to be assessed; 管理 δ is the adjustment factor for management performance and hazard management in relation to the target to be evaluated; 技术 The video AI and alarm analysis early warning coefficients are for the targets to be evaluated; α and β are preset process and site risk weight coefficients.

[0013] Furthermore, the inherent risk baseline value R 固有 The formula for determining it is:

[0014] ;

[0015] Among them, R 基准 The preset baseline value for the target to be evaluated; Let i be the risk amplification factor for the i-th hazard source included in the target to be evaluated. The status score is given to the i-th hazard source included in the target to be evaluated. The risk amplification factor is associated with the type of the corresponding hazard source, and the status score is associated with the aging degree and corrosion rate of the corresponding hazard source.

[0016] Process risk coefficient R 工艺 The formula for determining it is:

[0017] ;

[0018] in, This refers to the real-time temperature monitoring value of the target to be evaluated. Real-time pressure monitoring values ​​for the target to be evaluated; This refers to the real-time liquid level monitoring value for the target to be evaluated; This refers to the preset temperature standard value for the target to be evaluated; This refers to the preset pressure standard value for the target to be evaluated; This refers to the preset standard liquid level value for the target to be evaluated; The allowable temperature fluctuation range for the target to be evaluated; The allowable range of pressure fluctuations for the target to be evaluated; The permissible fluctuation range of the liquid level for the target to be evaluated;

[0019] The risk factor R of flammable / toxic gases in the workplace 场所 The formula for determining it is:

[0020] ;

[0021] ;

[0022] in, This refers to the real-time concentration of combustible gas at the workplace where the target to be evaluated is located. The LEL represents the real-time concentration of toxic gases in the workplace where the target is to be assessed; the IDL represents the minimum critical value at which the flammable gas in the workplace where the target is to be assessed has reached the explosion hazard concentration; and the IDLH represents the minimum critical value at which the toxic gas in the workplace where the target is to be assessed poses a threat to life and health.

[0023] Correction coefficient γ for management performance and hazard management 管理 The formula for determining it is:

[0024] ;

[0025] in, To assign responsibility weights to the targets to be evaluated, To assign a weight to the hazard management of the target to be evaluated, the weight of the responsibility of the supervisor is related to the frequency of safety inspections and the completeness of inspection records for the target to be evaluated, while the weight of hazard management is related to the duration of overdue rectification and the total number of hazards for the target to be evaluated.

[0026] Video AI and Alarm Analysis Early Warning Coefficient δ 技术 The formula for determining it is:

[0027] ;

[0028] ;

[0029] ;

[0030] ;

[0031] ;

[0032] The unit for delayed response time is minutes, and the total number of alarms is the sum of the number of valid warnings and the number of unassociated alarms.

[0033] Furthermore, the above method also includes: when the real-time concentration of the target gas in the target area is greater than or equal to a preset threshold, acquiring time-series data of the real-time concentration of the target gas in the target area over a target time period, wherein the target area is the area where any hazard source included in the target to be evaluated is located; determining the slope of change of the real-time concentration of the target gas in the target area based on the time-series data of the real-time concentration of the target gas in the target area over the target time period; determining the interference pattern corresponding to the slope of change of the real-time concentration of the target gas in the target area from a short-term spike interference feature library; wherein the short-term spike interference feature library stores the interference pattern corresponding to each of multiple slopes; when there is no slope of change of the real-time concentration of the target gas in the target area in the short-term spike interference feature library, incrementing the number of effective warnings by 1; when there is a slope of change of the real-time concentration of the target gas in the target area in the short-term spike interference feature library, incrementing the number of unassociated alarms by 1.

[0034] Furthermore, if the slope of the change in the real-time concentration of the target gas in the target area is not found in the short-term spike interference feature library, the number of valid warnings is incremented by 1. This includes: obtaining the real-time concentrations of the target gas in multiple areas associated with the target area when the slope of the change in the real-time concentration of the target gas in the target area is not found in the short-term spike interference feature library; determining the rate of change of the real-time concentration of the target gas in the multiple areas; incrementing the number of valid warnings by 1 when the number of areas with a rate of change of the real-time concentration of the target gas less than a preset rate of change threshold is greater than or equal to a preset number threshold; and incrementing the number of unassociated alarms by 1 when the number of areas with a rate of change of the real-time concentration of the target gas less than a preset rate of change threshold is less than a preset number threshold.

[0035] Furthermore, the above method also includes: when the real-time concentration of the target gas in the target area is greater than or equal to a preset threshold, acquiring process parameter information, microclimate information, and video data of the target area, the process parameter information including reactor temperature, pressure, and liquid level information, and the microclimate information including wind speed and temperature and humidity information; determining the leakage probability of the target gas based on a pre-trained Bayesian neural network; incrementing the number of effective warnings by 1 when the leakage probability of the target gas is greater than or equal to a preset probability threshold; and incrementing the number of unassociated alarms by 1 when the leakage probability of the target gas is less than the preset probability threshold.

[0036] Furthermore, the above method also includes: in the case of a major risk level, activating the corresponding material allocation system according to the emergency response plan corresponding to a major risk level, sending a first alarm message to multiple terminal devices, the first alarm message being used to indicate that the risk level is major risk, and establishing communication with multiple terminal devices to realize information transmission between multiple terminal devices; in the case of a relatively high risk, a general risk, or a low risk level, sending a second alarm message to multiple terminal devices according to the emergency response plan corresponding to a relatively high risk, a general risk, or a low risk level, the second alarm message being used to indicate that the risk level is relatively high risk, a general risk, or a low risk.

[0037] Furthermore, the above method also includes: when the risk level of the target to be assessed is major risk, a first indicator is displayed on the preset interface, and the first indicator is red; when the risk level of the target to be assessed is relatively high risk, a second indicator is displayed on the preset interface, and the second indicator is orange; when the risk level of the target to be assessed is moderate risk, a third indicator is displayed on the preset interface, and the third indicator is yellow; when the risk level of the target to be assessed is low risk, a fourth indicator is displayed on the preset interface, and the fourth indicator is blue.

[0038] The beneficial effects of this invention are:

[0039] By acquiring multimodal monitoring data of the target to be assessed (major hazard source), and determining the comprehensive risk value of the target based on the multimodal monitoring data; and then determining the risk level of the target based on the comprehensive risk value, the risk level of the major hazard source can be quickly and accurately determined. This enables real-time assessment, analysis, and display of the safety risks of major hazard sources, and supports the automatic sending, verification, feedback, and supervision of early warning information based on the warning level. In other words, the method provided by this invention can quickly and accurately assess the risks of major hazard sources, improving accuracy and timeliness, and meeting the risk assessment needs of users in different usage scenarios.

[0040] Secondly, the present invention provides a risk assessment system for major hazard sources based on multimodal analysis, comprising:

[0041] The data acquisition module is used to acquire monitoring data of the target to be evaluated. The target to be evaluated includes multiple hazard sources, including at least one of storage tanks, equipment, and hazardous chemical warehouses. The monitoring data includes real-time monitoring data and alarm data of the corresponding hazard source, such as liquid level, temperature, pressure data, and concentration of flammable and toxic gases.

[0042] The risk value determination module is used to dynamically adjust the process production risk coefficient, workplace combustible / toxic gas risk coefficient, management performance and hazard management correction coefficient, and video AI and alarm analysis early warning coefficient based on the fluctuation of the monitoring data of the target to be evaluated. The process production risk coefficient is used to characterize the degree of risk existing in the process production of the hazard source; the workplace combustible / toxic gas risk coefficient is used to characterize the degree of risk of combustible / toxic gases in the workplace of the hazard source; the management performance and hazard management correction coefficient is used to characterize the degree of impact of management performance and hazard management on the risk existing in the hazard source; and the video AI and alarm analysis early warning coefficient is used to characterize the degree of risk existing in the video data of the hazard source.

[0043] The risk value determination module is also used to determine the comprehensive risk value of the target to be evaluated based on the risk coefficient of the production process, the risk coefficient of flammable / toxic gases in the workplace, the correction coefficient of management performance and hazard control, and the early warning coefficient of video AI and alarm analysis.

[0044] The risk level determination module is used to determine the risk level of the target to be evaluated based on its comprehensive risk value. Different risk levels correspond to different ranges of comprehensive risk values.

[0045] Thirdly, the present invention provides an electronic device, comprising: a memory and one or more processors; the memory and the processors are coupled; wherein the memory stores computer program code, the computer program code including computer instructions, which, when executed by the processor, cause the electronic device to perform the personnel risk assessment method of any of the first aspects described above.

[0046] Fourthly, a computer-readable storage medium is provided, including computer instructions that, when executed on an electronic device, cause the electronic device to perform the personnel risk assessment method described in any of the first aspects above.

[0047] Fifthly, a computer program product is provided that, when run on a computer, causes the computer to perform the personnel risk assessment method described in any of the first aspects above.

[0048] Understandably, the beneficial effects of the personnel risk assessment system of the second aspect, the electronic equipment of the third aspect, the computer-readable storage medium of the fourth aspect, and the computer program product of the fifth aspect provided above can be referred to the beneficial effects of the first aspect and any of its possible design methods, and will not be repeated here. Attached Figure Description

[0049] Figure 1 A flowchart illustrating a method for risk assessment of major hazard sources based on multimodal analysis provided by this invention;

[0050] Figure 2 A flowchart illustrating another method for risk assessment of major hazard sources based on multimodal analysis provided by the present invention;

[0051] Figure 3 This is a schematic diagram of a major hazard source risk assessment system provided by the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. In the description of this application, unless otherwise stated, " / " indicates that the objects before and after are in an "or" relationship. For example, A / B can represent A or B. "And / or" in this application is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Furthermore, in the description of this application, unless otherwise stated, "multiple" refers to two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple. Furthermore, to facilitate a clear description of the technical solutions in the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with substantially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that "first" and "second" are not necessarily different. Meanwhile, in the embodiments of this application, words such as "exemplary" or "for example" are used to indicate that something is being used as an example, illustration, or description. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes.

[0053] Current risk management and early warning technologies for major hazard sources are gradually shifting towards intelligent and digital approaches. Existing technologies primarily utilize IoT sensor networks to collect key parameters such as temperature, pressure, and leakage concentration in real time, combining this with big data analysis to construct risk assessment indicator systems (such as the Analytic Hierarchy Process (AHP) and fuzzy comprehensive evaluation). Some systems incorporate machine learning algorithms (such as LSTM and random forest) to mine features from historical accident data. Regarding early warning mechanisms, most solutions employ multi-level threshold-triggered alarm modes, and some advanced systems have integrated GIS visualization platforms and emergency response databases. In recent years, the fusion of multi-source heterogeneous data (such as equipment status data, meteorological and environmental data, and video surveillance data) and the optimization of dynamic risk assessment models have become research hotspots.

[0054] However, the following shortcomings still exist in the relevant technologies: Insufficient data integration and real-time performance: Traditional methods rely on single sensors or manual inspections, and lack an efficient fusion mechanism for multi-source heterogeneous data (such as equipment status, environmental parameters, and video surveillance), resulting in incomplete risk feature extraction and high data update delays, making it difficult to capture dynamic risks in a timely manner; Furthermore, the early warning models in the relevant technologies have weak generalization capabilities, poor adaptability to complex scenarios (such as multi-hazard coupling and equipment nonlinear failure), high false alarm / false alarm rates, and model training relies on historical data, making it difficult to cope with sudden abnormal operating conditions; In addition, the relevant technologies only assess a single hazard independently, without fully considering the dynamic correlation of risks from multiple dimensions such as equipment, environment, and human operation, resulting in insufficient accuracy in global risk prediction; Moreover, they lack self-learning and optimization capabilities, cannot adjust risk weights according to real-time operating conditions, and the early warning response strategies are fixed, making it difficult to adapt to long-term evolving risks such as changes in production processes or equipment aging. Therefore, there are problems with poor accuracy and timeliness in the risk assessment methods for major hazard sources.

[0055] Therefore, there is an urgent need for a risk assessment method and system for major hazard sources based on multimodal analysis, which can quickly and accurately assess the risks of major hazard sources, improve accuracy and timeliness, and meet the risk assessment needs of users in different usage scenarios.

[0056] For the above issues, please refer to Figure 1 This invention provides a method for risk assessment of major hazard sources based on multimodal analysis, including steps S101-S104:

[0057] S101: Obtain monitoring data for the target to be evaluated.

[0058] Specifically, the targets to be assessed include multiple hazard sources, including at least one of storage tanks, equipment, and hazardous chemical warehouses. The monitoring data includes real-time monitoring data and alarm data of the corresponding hazard source's liquid level, temperature, pressure, and concentration of flammable and toxic gases.

[0059] It should be understood that the target to be evaluated is a major hazard source that includes multiple hazard sources. Multiple hazard sources can be classified into one major hazard source based on the area where they are located, multiple hazard sources can be classified into one major hazard source based on different parts of the process flow, or multiple hazard sources can be classified into one major hazard source based on the project to which they belong. The embodiments of the present invention do not impose any special restrictions on the specific classification method of major hazard sources.

[0060] In one example, the specific implementation of acquiring monitoring data for the target to be evaluated is as follows: multimodal data collaborative acquisition, access to heterogeneous sensor data (including liquid level / pressure / temperature sensors, toxic and flammable gas concentration detectors) and video surveillance information, and data normalization preprocessing through edge computing nodes to eliminate data heterogeneity and latency differences. This can also be understood as: the monitoring data consists of sensor monitoring data from the major hazard source (the target to be evaluated), including: liquid level, pressure, temperature, flammable gas detection, toxic gas detection, real-time monitoring data and alarm data of liquid level, temperature, pressure, and flammable and toxic gas concentrations at storage tanks, equipment, and hazardous chemical warehouses, querying historical data and comparative analysis to provide data support for online safety spot checks of major hazard sources. The alarm data consists of video surveillance footage of major hazard sources, including intelligent analysis of monitoring videos from key areas such as ammonium nitrate warehouses, central control rooms, and major hazard source sites, providing data support for comprehensive identification and early warning of fires, smoke, and personnel violations (absence from control room staff).

[0061] In one possible implementation, the method provided by the embodiments of the present invention further includes:

[0062] The monitoring data for the targets to be evaluated are cleaned and sorted to ensure the integrity and consistency of the data.

[0063] Specifically, it integrates monitoring data from major hazard sources, including basic information on major hazard sources, AI information from surveillance videos, sensor monitoring data, responsibility performance data, and hazard investigation and management data, and presents the results to end users. It possesses cross-cloud, cross-network, and cross-data center data integration capabilities, providing application programming interfaces / gateways, supporting message publishing and subscription, multi-cluster deployment, and message trajectory tracking. Based on data integration, it performs various data mining and analysis methods, including descriptive analysis, diagnostic analysis, predictive analysis, and causal analysis; it provides statistical analysis, retrieval, machine learning, text analysis, video analysis, and other analytical methods, models, and tools to help optimize the construction and management of the industrial internet intelligent supervision platform.

[0064] It should be noted that basic information on major hazard sources refers to the basic data and information describing and identifying major hazard sources. This information is crucial for assessing, monitoring, and managing major hazard sources. Basic information on major hazard sources typically includes the following aspects: a basic description of the hazard source (including the name and location of the hazard source, the types, quantities, and characteristics of the hazardous substances involved, and the scale and method of production, storage, use, or operation of the hazard source), the types of accidents that may occur with the hazard source, the inherent risk level and potential impact of the hazard source, emergency response and accident handling plans and measures, and relevant laws, regulations, and standards for hazard source management.

[0065] AI-powered video surveillance of major hazard sources refers to the use of artificial intelligence technology to analyze and process video data from major hazard sources in real time, enabling intelligent monitoring, early warning, and response. This technology combines computer vision, machine learning, deep learning, and other AI technologies to automatically identify and analyze anomalies in videos, such as smoke, fires, employees sleeping on duty, and employees leaving their posts, thereby improving the accuracy and efficiency of monitoring.

[0066] Sensor monitoring of major hazard sources refers to the real-time monitoring of key parameters at the site of major hazard sources using various types of sensors. Appropriate sensor types are selected according to monitoring needs, such as temperature sensors, pressure sensors, liquid level sensors, gas sensors (detecting combustible and toxic gases), vibration sensors, flow sensors, etc.

[0067] Sensor monitoring is an important component of safety management for major hazard sources. It provides real-time and accurate data to help managers understand the status of hazard sources and take necessary preventive measures.

[0068] The responsibility for ensuring the safety of major hazard sources refers to the comprehensive and continuous supervision and management of major hazard sources by key personnel such as the company's principal responsible person, technical responsible person, and operation responsible person, in order to ensure the safety of major hazard sources and to ensure that the safety risks of major hazard sources are effectively controlled and accidents are prevented.

[0069] The responsibility for ensuring the safety of major hazard sources includes the following:

[0070] Establish and improve a comprehensive safety responsibility system for major hazard sources, clearly defining the safety responsibilities of personnel at all levels. Develop and improve safety management systems and operating procedures for major hazard sources, ensuring their effective implementation. Conduct regular safety inspections and assessments of major hazard sources to promptly identify and eliminate potential safety hazards. Develop and implement emergency response plans for major hazard sources to ensure timely and effective handling and rescue in the event of an accident. Strengthen safety training and education for major hazard sources to improve employees' safety awareness and operational skills. Regularly organize drills for major hazard sources to test the effectiveness of emergency response plans and employees' emergency response capabilities. Conduct regular summaries and assessments of the safety management of major hazard sources to continuously improve and enhance safety management effectiveness.

[0071] The investigation and management of major hazard sources refers to a comprehensive and systematic inspection of major hazard sources, identification of existing safety hazards, and the implementation of effective measures to rectify them in order to eliminate or reduce safety hazards and ensure the safe operation of major hazard sources.

[0072] The investigation and rectification of major hazard sources includes the following:

[0073] Develop a hazard identification plan: Based on the characteristics, scale, and potential risks of major hazard sources, develop a detailed hazard identification plan, clearly defining the scope, content, methods, and timeline of the identification. Conduct hazard identification: In accordance with the hazard identification plan, organize professional personnel to conduct a comprehensive and systematic inspection of major hazard sources, including equipment and facilities, technological processes, safety protection measures, and emergency rescue facilities. Identify safety hazards: During the hazard identification process, identify existing safety hazards through observation, detection, and testing, including defects in equipment and facilities, unreasonable technological processes, and insufficient safety protection measures. Analyze safety hazards: Conduct in-depth analysis of the identified safety hazards to determine their causes, scope of impact, and severity, and assess their risk to the safe operation of major hazard sources. Develop rectification plans: Based on the analysis results of safety hazards, develop corresponding rectification plans, clearly defining the rectification objectives, measures, responsibilities, and time requirements. Implement hazard rectification: In accordance with the rectification plan, organize personnel to rectify the safety hazards, including the repair of equipment and facilities, optimization of technological processes, and improvement of safety protection measures. Verify the effectiveness of rectification: Verify the rectified major hazard sources to ensure that safety hazards have been effectively eliminated or reduced, meeting the requirements for safe operation. Establish hazard investigation and management files: Record and archive the process, results, and rectification status of hazard investigation and management, serving as an important basis for the safety management of major hazard sources.

[0074] S102: Determine the process risk coefficient, workplace combustible / toxic gas risk coefficient, management performance and hazard mitigation correction coefficient, and video AI and alarm analysis early warning coefficient based on the monitoring data of the target to be evaluated.

[0075] Specifically, the process production risk coefficient is used to characterize the degree of risk present in the process production of the hazard source; the workplace combustible / toxic gas risk coefficient is used to characterize the degree of risk of combustible / toxic gases in the workplace of the hazard source; the management performance and hazard management correction coefficient is used to characterize the degree of impact of management performance and hazard management on the risks present in the hazard source; and the video AI and alarm analysis early warning coefficient is used to characterize the degree of risk present in the video data of the hazard source.

[0076] Process risk coefficient R 工艺 The formula for determining it is:

[0077] ;

[0078] in, This is a real-time temperature monitoring value; This is a real-time pressure monitoring value; This is the real-time liquid level monitoring value; This is the preset temperature standard value; The preset pressure standard value; The preset liquid level standard value; This refers to the allowable temperature fluctuation range. This refers to the allowable pressure fluctuation range; This refers to the allowable fluctuation range of the liquid level.

[0079] In one example .

[0080] It should be noted that R exceeds the threshold. 工艺 Based on 100%, dynamic correction is triggered.

[0081] The risk factor R of flammable / toxic gases in the workplace 场所 The formula for determining it is:

[0082] ;

[0083] ;

[0084] in, This represents the real-time concentration of combustible gas. The LEL represents the real-time concentration of toxic gases; LEL represents the minimum critical value at which combustible gases reach the explosion hazard concentration; IDLH represents the minimum critical value at which toxic gases threaten life and health.

[0085] It should be noted that LEL (Less than Explosive Altitude) is the minimum concentration of a flammable gas or vapor in air. When this concentration is reached or exceeded, it may cause an explosion or combustion upon contact with an ignition source (spark, high temperature, etc.). The unit is usually expressed as a volume percentage (%vol), for example, the LEL for methane is 5% (meaning that when methane accounts for ≥5% of the air, there is an explosion risk). LEL is the minimum critical value for determining whether a flammable gas has reached an explosive hazard concentration.

[0086] IDLH stands for Immediately Life-Threatening Health Concentration. It refers to the concentration level of a toxic substance in the air at which short-term exposure (usually ≤30 minutes) can lead to irreversible health damage, inability to escape, or death. The unit is ppm (parts per million) or mg / m³, for example, the IDLH of chlorine is 10 ppm. IDLH is an important indicator for determining whether an emergency escape respirator (EAR) is necessary.

[0087] Correction coefficient γ for management performance and hazard management 管理 The formula for determining it is:

[0088] ;

[0089] in, To ensure the effective performance of duties, The weighting of hazard management is related to the frequency of safety inspections and the completeness of inspection records, while the weighting of hazard management is related to the duration of overdue rectification and the total number of hazards.

[0090] Specifically, .

[0091] Video AI and Alarm Analysis Early Warning Coefficient δ 技术 The formula for determining it is:

[0092] ;

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] The unit for delayed response time is minutes, and the total number of alarms is the sum of the number of valid warnings and the number of unassociated alarms.

[0098] The method provided in this invention dynamically couples multiple coefficients. Through... The invention addresses the multiple impacts of equipment operation, human factors, and technology, resolving issues such as insufficient data integration and real-time performance, lack of risk coupling analysis, and low levels of adaptability and intelligence inherent in traditional algorithms. Furthermore, exceeding limits for any indicator (such as toxic gas) directly escalates the warning level, while also considering the correlated effects of warnings, resulting in more diverse and reasonable risk assessments and overcoming the weak generalization ability of traditional warning models. Finally, the method provided in this embodiment supports adjusting weights... It can be adapted to different industry scenarios and meet the user's needs in different usage scenarios.

[0099] S103: Determine the comprehensive risk value of the target to be evaluated based on the risk coefficient of the production process, the risk coefficient of flammable / toxic gases in the workplace, the correction coefficient of management performance and hazard control, and the early warning coefficient of video AI and alarm analysis;

[0100] In some embodiments, the formula for determining the comprehensive risk value R is:

[0101] ；

[0102] Among them, R 固有 R represents the inherent risk baseline of the target to be evaluated. 工艺 R represents the risk coefficient of the target to be evaluated in the production process. 场所 γ represents the flammable / toxic gas risk factor of the workplace where the target is to be assessed; 管理 δ is the adjustment factor for management performance and hazard management in relation to the target to be evaluated; 技术 The video AI and alarm analysis early warning coefficients are for the targets to be evaluated; α and β are preset process and site risk weight coefficients.

[0103] In one example, α=0.6, β=0.4.

[0104] Furthermore, the inherent risk baseline value R 固有 The formula for determining it is:

[0105] ;

[0106] Among them, R 基准 The preset baseline value for the target to be evaluated; Let i be the risk amplification factor for the i-th hazard source included in the target to be evaluated. The status score is given to the i-th hazard source included in the target to be evaluated. The risk amplification factor is associated with the type of the corresponding hazard source, and the status score is associated with the aging degree and corrosion rate of the corresponding hazard source.

[0107] In one example, the hazardous source toxic substance k1=0.2, and the hazardous source high-voltage equipment k2=0.15.

[0108] In some embodiments, the method provided by the present invention further includes:

[0109] If the real-time concentration of the target gas in the target area is greater than or equal to a preset threshold, time-series data of the real-time concentration of the target gas in the target area over a target time period is obtained. The target area is the area where any hazard source included in the target to be evaluated is located. The slope of the change in the real-time concentration of the target gas in the target area is determined based on the time-series data of the real-time concentration of the target gas in the target area over the target time period. The interference pattern corresponding to the slope of the change in the real-time concentration of the target gas in the target area is determined from the short-term spike interference feature library. The short-term spike interference feature library stores the interference pattern corresponding to each of the multiple slopes. If the slope of the change in the real-time concentration of the target gas in the target area is not found in the short-term spike interference feature library, the number of valid warnings is incremented by 1. If the slope of the change in the real-time concentration of the target gas in the target area is found in the short-term spike interference feature library, the number of unassociated alarms is incremented by 1.

[0110] Furthermore, if the slope of the real-time concentration change of the target gas in the target area is not found in the short-term spike interference feature library, the number of effective warnings will be incremented by 1, including:

[0111] If the slope of the change in the real-time concentration of the target gas in the target area is not found in the short-term spike interference feature library, the real-time concentration of the target gas in multiple areas associated with the target area is obtained; the rate of change of the real-time concentration of the target gas in multiple areas is determined; if the number of areas where the rate of change of the real-time concentration of the target gas is less than the preset rate of change threshold is greater than or equal to the preset number threshold, the number of valid warnings is incremented by 1; if the number of areas where the rate of change of the real-time concentration of the target gas is less than the preset rate of change threshold is less than the preset number threshold, the number of unassociated alarms is incremented by 1.

[0112] In other embodiments, the method provided by the present invention further includes:

[0113] If the real-time concentration of the target gas in the target area is greater than or equal to a preset threshold, process parameter information, microclimate information, and video data of the target area are acquired. The process parameter information includes reactor temperature, pressure, and liquid level information, and the microclimate information includes wind speed, temperature, and humidity information. The leakage probability of the target gas is determined based on a pre-trained Bayesian neural network. If the leakage probability of the target gas is greater than or equal to a preset probability threshold, the number of effective warnings is incremented by 1. If the leakage probability of the target gas is less than the preset probability threshold, the number of unassociated alarms is incremented by 1.

[0114] As can be seen from the above, the method provided in this embodiment of the invention addresses the problem of instantaneous false alarms in the monitoring of toxic and flammable gases in major hazard source areas by constructing a multi-dimensional intelligent identification system: employing two filtering mechanisms for false alarm determination:

[0115] Firstly, based on preliminary screening using time-series pattern recognition, a sliding time window algorithm is used to analyze the deviation of the concentration value's abrupt change slope and duration from the GB / T 50493 standard threshold, establishing a short-term spike interference feature library. Based on this library, the system accurately determines whether an alarm is valid. Furthermore, when a single-point alarm occurs (target area), the system automatically verifies the concentration change trend of adjacent monitoring units (within a 50-meter radius). If no spatial correlation gradient is formed, a false alarm flag is triggered. Alternatively, multi-system data coupling analysis is performed, linking the enterprise's DCS system to obtain process parameters (such as reactor temperature, pressure, and liquid level), meteorological station microclimate data (wind speed, temperature, and humidity), and video surveillance intelligent recognition results. A Bayesian network is then used to construct a causal inference model to calculate the true leakage probability value.

[0116] Through actual testing, this multi-dimensional intelligent identification system has reduced the false alarm rate by more than 90%. At the same time, it has established a self-learning optimization module, which extracts features from historical false alarm cases through the platform and continuously updates the dynamic compensation coefficient of device sensitivity and the weight library of environmental interference factors.

[0117] S104: Determine the risk level of the target to be evaluated based on its comprehensive risk value, where different risk levels correspond to different ranges of comprehensive risk values.

[0118] In some embodiments, see Figure 2 The risk levels include major risk, significant risk, general risk, and low risk; the above S104 includes:

[0119] S1041: If the overall risk value of the target to be assessed is greater than or equal to the first risk value (e.g., 61), the risk level of the target to be assessed shall be determined as a major risk.

[0120] S1042: If the overall risk value of the target to be assessed is greater than or equal to the second risk value (e.g., 42) and less than the first risk value (e.g., 61), the risk level of the target to be assessed is determined to be a greater risk.

[0121] S1043: If the overall risk value of the target to be assessed is greater than or equal to the third risk value (e.g., 21) and less than the second risk value (e.g., 61), the risk level of the target to be assessed is determined to be general risk.

[0122] S1044: If the overall risk value of the target to be evaluated is less than the third risk value (e.g., 21), the risk level of the target to be evaluated is determined to be low risk.

[0123] In some embodiments, the present invention further includes:

[0124] In the event of a major risk level, the corresponding material allocation system is activated according to the emergency response plan for a major risk level, and the first alarm information is sent to multiple terminal devices. The first alarm information is used to indicate that the risk level is major risk and to establish communication with multiple terminal devices to realize information transmission between multiple terminal devices.

[0125] When the risk level is relatively high, moderate, or low, a second alarm message is sent to multiple terminal devices according to the emergency response plan corresponding to the risk level of relatively high, moderate, or low risk. The second alarm message is used to indicate whether the risk level is relatively high, moderate, or low.

[0126] It can also be understood that: when the risk level is major risk, the method provided by this invention will, after completing the accurate matching of the emergency plan, simultaneously activate the material allocation system, communication support module and expert consultation mechanism associated with the plan, forming a closed-loop management of the entire process of "monitoring and early warning - intelligent matching - plan generation - collaborative disposal", which significantly improves the timeliness of emergency response and the scientific nature of disposal plan.

[0127] Furthermore, the method provided in this embodiment of the invention requires coordinated action when the risk level is low, moderate, or significant. The platform will automatically notify relevant responsible persons in the enterprise, including the safety manager and safety officer.

[0128] In some embodiments, the present invention further includes:

[0129] When the risk level is "major risk", the first indicator will be displayed on the preset interface in red; when the risk level is "relatively high risk", the second indicator will be displayed on the preset interface in orange; when the risk level is "moderate risk", the third indicator will be displayed on the preset interface in yellow; and when the risk level is "low risk", the fourth indicator will be displayed on the preset interface in blue.

[0130] As described in S101-S104 above, the method provided by this embodiment of the invention acquires multimodal monitoring data of the target to be evaluated and determines the comprehensive risk value of the target based on the multimodal monitoring data; then, it determines the risk level of the target based on the comprehensive risk value. This method can quickly and accurately determine the risk level of major hazard sources, realize real-time assessment, analysis, and display of safety risks of major hazard sources, and support the immediate and automatic sending, verification, feedback, and supervision of early warning information based on the early warning level. In other words, the method provided by this invention can quickly and accurately achieve risk assessment of major hazard sources, improve accuracy and timeliness, and meet the risk assessment needs of users in different usage scenarios.

[0131] See Figure 3 The present invention also provides a risk assessment system for major hazard sources based on multimodal analysis, comprising: a data acquisition module for acquiring monitoring data of the target to be assessed; the target to be assessed includes multiple hazard sources, including storage tanks, devices or hazardous chemical warehouses, and the monitoring data includes real-time monitoring data and alarm data of the liquid level, temperature, pressure, and concentration of flammable and toxic gases of the corresponding hazard sources;

[0132] The risk value determination module is used to determine the process production risk coefficient, workplace combustible / toxic gas risk coefficient, management performance and hazard management correction coefficient, and video AI and alarm analysis early warning coefficient based on the monitoring data of the target to be evaluated. The process production risk coefficient is used to characterize the degree of risk present in the process production of the hazard source; the workplace combustible / toxic gas risk coefficient is used to characterize the degree of risk of combustible / toxic gases in the workplace of the hazard source; the management performance and hazard management correction coefficient is used to characterize the degree of impact of management performance and hazard management on the risk present in the hazard source; and the video AI and alarm analysis early warning coefficient is used to characterize the degree of risk present in the video data of the hazard source. The risk value determination module is also used to determine the comprehensive risk value of the target to be evaluated based on the process production risk coefficient, workplace combustible / toxic gas risk coefficient, management performance and hazard management correction coefficient, and video AI and alarm analysis early warning coefficient.

[0133] The risk level determination module is used to determine the risk level of the target to be evaluated based on its comprehensive risk value. Different risk levels correspond to different ranges of comprehensive risk values.

[0134] In some solutions, multiple embodiments of this application can be combined, and the combined solution can be implemented. Optionally, some operations in the processes of each method embodiment may be combined, and / or the order of some operations may be changed. Furthermore, the execution order between the steps of each process is merely exemplary and does not constitute a limitation on the execution order between steps; other execution orders are also possible. It is not intended to indicate that the execution order is the only possible order in which these operations can be performed. Those skilled in the art will conceive of various ways to reorder the operations described herein. In addition, it should be noted that the process details involved in one embodiment of this document are similarly applicable to other embodiments, or different embodiments may be combined.

[0135] Furthermore, some steps in the method embodiments can be equivalently replaced with other possible steps. Alternatively, some steps in the method embodiments may be optional and can be deleted in certain use cases. Or, other possible steps may be added to the method embodiments. Moreover, the various method embodiments can be implemented individually or in combination.

[0136] Through the above description of the implementation methods, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above.

[0137] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of systems or units may be electrical, mechanical, or other forms.

[0138] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0139] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to the solution, or all or part of the technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0140] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A major hazard source risk assessment method based on multi-modal analysis, characterized in that, The method comprises: obtaining monitoring data of a target to be evaluated, wherein the target to be evaluated comprises a plurality of hazard sources, the hazard sources comprise at least one of a storage tank, a device, and a hazardous chemical warehouse, and the monitoring data comprises real-time monitoring data and alarm data of a liquid level, a temperature, a pressure, and a concentration of a flammable and / or toxic gas corresponding to the hazard sources; determining a process production process risk coefficient, an operation site flammable / toxic gas risk coefficient, a management performance and hidden danger treatment correction coefficient, and a video AI and alarm analysis early warning coefficient according to the monitoring data of the target to be evaluated, wherein the process production process risk coefficient is used to represent a risk degree existing in a process production process of the hazard source, the operation site flammable / toxic gas risk coefficient is used to represent a risk degree of a flammable / toxic gas in an operation site of the hazard source, the management performance and hidden danger treatment correction coefficient is used to represent an influence degree of management performance and hidden danger treatment on the risk existing in the hazard source, and the video AI and alarm analysis early warning coefficient is used to represent a risk degree existing in video data of the hazard source; determining a comprehensive risk value of the target to be evaluated according to the process production process risk coefficient, the operation site flammable / toxic gas risk coefficient, the management performance and hidden danger treatment correction coefficient, and the video AI and alarm analysis early warning coefficient; determining a risk level of the target to be evaluated according to the comprehensive risk value of the target to be evaluated, wherein different risk levels correspond to different numerical intervals of the comprehensive risk value; wherein a determination formula of the comprehensive risk value R is: R = R 固有 × (1 + a · R 工艺 + b · R 场所 ) × g 管理 × d 技术 ; wherein, R 固有 is the inherent risk base value of the target to be evaluated; R 工艺 is the risk coefficient of the target to be evaluated in the process production process; R 场所 is the flammable / toxic gas risk coefficient of the work site where the target to be evaluated is located; γ 管理 is the management duty and hidden danger management correction coefficient for the target to be evaluated; δ 技术 is the video AI and alarm analysis early warning coefficient for the target to be evaluated; α is a preset process risk weight coefficient, and β is a preset site risk weight coefficient; The intrinsic risk base value The determination formula is: ； wherein R 基准 is a preset reference value of the target to be evaluated; is a risk amplification factor of an i-th hazard source included in the target to be evaluated, is a state score of the i-th hazard source included in the target to be evaluated, the risk amplification factor being associated with a kind of the corresponding hazard source, and the state score being associated with an aging degree and a corrosion rate of the corresponding hazard source. The process production risk coefficient R 工艺 The determination formula is: ； wherein, is a real-time temperature monitoring value for the target to be evaluated; is a real-time pressure monitoring value for the target to be evaluated; is a real-time liquid level monitoring value for the target to be evaluated; is a preset temperature standard value for the target to be evaluated; is a preset pressure standard value for the target to be evaluated; is a preset liquid level standard value for the target to be evaluated; is a temperature allowable fluctuation range for the target to be evaluated; is a pressure allowable fluctuation range for the target to be evaluated; is a liquid level allowable fluctuation range for the target to be evaluated; The work site flammable / toxic gas risk coefficient R 场所 The determination formula is: ； ； wherein, a real-time concentration of combustible gas in the working site where the target to be evaluated is located, a real-time concentration of toxic gas in the working site where the target to be evaluated is located, LEL is the minimum critical value of the combustible gas reaching the explosive dangerous concentration in the working site where the target to be evaluated is located; IDLH is the minimum critical value of the toxic gas threatening the life and health concentration in the working site where the target to be evaluated is located. The management performance and hidden danger management correction coefficient γ 管理 The determination formula is: ； wherein, is a package performance weight for the target to be evaluated, is a hidden danger management weight for the target to be evaluated, the package performance weight is associated with a safety inspection frequency and an inspection record integrity for the target to be evaluated, and the hidden danger management weight is associated with an overdue unrectification duration and a total number of hidden dangers for the target to be evaluated. The video AI and the alarm analysis early warning coefficient δ 技术 The determination formula is: ； ； ； ； ； wherein a unit of the delay response time is min, and the total number of alarms is a sum of the effective early warning times and the unassociated alarm number.

2. The method of claim 1, wherein, The risk level comprises a major risk, a relatively major risk, a general risk, and a low risk. The determination of the risk level of the target to be evaluated according to the comprehensive risk value of the target to be evaluated comprises: in a case where the comprehensive risk value of the target to be evaluated is greater than or equal to a first risk value, determining that the risk level of the target to be evaluated is the major risk; in a case where the comprehensive risk value of the target to be evaluated is greater than or equal to a second risk value and less than the first risk value, determining that the risk level of the target to be evaluated is the relatively major risk; in a case where the comprehensive risk value of the target to be evaluated is greater than or equal to a third risk value and less than the second risk value, determining that the risk level of the target to be evaluated is the general risk; in a case where the comprehensive risk value of the target to be evaluated is less than the third risk value, determining that the risk level of the target to be evaluated is the low risk.

3. The method of claim 2, wherein, The method further comprises: in a case where a real-time concentration of a target gas in a target area is greater than or equal to a preset threshold value, obtaining time series data of the real-time concentration of the target gas in the target area in a target time period, wherein the target area is an area where any hazard source comprised in the target to be evaluated is located; determining a change slope of the real-time concentration of the target gas in the target area according to the time series data of the real-time concentration of the target gas in the target area in the target time period. determine, from a short-time peak interference feature library, an interference mode corresponding to a change slope of the real-time concentration of the target gas in the target region, wherein the short-time peak interference feature library stores an interference mode corresponding to each of a plurality of change slopes; in a case where the change slope of the real-time concentration of the target gas in the target region does not exist in the short-time peak interference feature library, add 1 to the number of effective early warnings; in a case where the change slope of the real-time concentration of the target gas in the target region exists in the short-time peak interference feature library, add 1 to the number of unassociated alarms.

4. The method of claim 3, wherein, The method further comprises: in a case where the real-time concentration of the target gas in the target region is greater than or equal to a preset threshold, obtaining process parameter information, microclimate information, and video data of the target region, wherein the process parameter information includes reaction kettle temperature, pressure, and liquid level information, and the microclimate information includes wind speed and temperature and humidity information; determining a leakage probability of the target gas based on a pre-trained Bayesian neural network; in a case where the leakage probability of the target gas is greater than or equal to a preset probability threshold, add 1 to the number of effective early warnings; in a case where the leakage probability of the target gas is less than the preset probability threshold, add 1 to the number of unassociated alarms.

5. The method of claim 4, wherein, The method further comprises: in a case where the risk level is a major risk, starting a corresponding material allocation system according to an emergency handling plan corresponding to the risk level being a major risk, sending first alarm information to a plurality of terminal devices, the first alarm information being used to prompt that the risk level is a major risk, and establishing communication with the plurality of terminal devices to realize information transmission between the plurality of terminal devices; in a case where the risk level is a major risk, a general risk, or a low risk, sending second alarm information to a plurality of terminal devices according to an emergency handling plan corresponding to the risk level being a major risk, a general risk, or a low risk, the second alarm information being used to prompt that the risk level is a major risk, a general risk, or a low risk. The method further comprises: in a case where the risk level of the target to be evaluated is a major risk, displaying a first identifier on a preset interface, the first identifier being red; 6. The method of claim 5, wherein, in a case where the risk level of the target to be evaluated is a major risk, displaying a second identifier on a preset interface, the second identifier being orange; ​ ​ 7. The method of claim 6, wherein, ​ ​ ​ In a case where the risk level of the target to be evaluated is a general risk, a third identifier is displayed on a preset interface, the third identifier being yellow; In a case where the risk level of the target to be evaluated is a low risk, a fourth identifier is displayed on a preset interface, the fourth identifier being blue.

8. A major hazard source risk assessment system based on multi-modal analysis, characterized in that, Comprise: a data acquisition module configured to acquire monitoring data of a target to be evaluated; the target to be evaluated comprises a plurality of hazard sources, the hazard sources comprising at least one of a storage tank, a device, and a hazardous chemical warehouse, and the monitoring data comprises real-time monitoring data and alarm data of a liquid level, a temperature, a pressure data, and a combustible and toxic gas concentration corresponding to the hazard sources; a risk value determination module configured to determine a process production process risk coefficient, an operation site combustible / toxic gas risk coefficient, a management performance and hidden danger treatment correction coefficient, and a video AI and alarm analysis early warning coefficient according to the monitoring data of the target to be evaluated; wherein the process production process risk coefficient is used to represent the risk degree existing in the process production process of the hazard source, the operation site combustible / toxic gas risk coefficient is used to represent the risk degree of the combustible / toxic gas in the operation site of the hazard source, the management performance and hidden danger treatment correction coefficient is used to represent the influence degree of the management performance and hidden danger treatment on the risk existing in the hazard source, and the video AI and alarm analysis early warning coefficient is used to represent the risk degree existing in the video data of the hazard source; The risk value determination module is further configured to determine a comprehensive risk value of the target to be evaluated according to the process production process risk coefficient, the operation site combustible / toxic gas risk coefficient, the management performance and hidden danger treatment correction coefficient, and the video AI and alarm analysis early warning coefficient; a risk level determination module configured to determine a risk level of the target to be evaluated according to the comprehensive risk value of the target to be evaluated, wherein different risk levels correspond to different numerical intervals of the comprehensive risk value respectively; wherein the determination formula of the comprehensive risk value R is: ； wherein, R 固有 is the inherent risk base value of the target to be evaluated; R 工艺 is the risk coefficient of the target to be evaluated in the process production; R 场所 is the flammable / toxic gas risk coefficient of the work site where the target to be evaluated is located; γ 管理 is the management duty and hidden danger management correction coefficient for the target to be evaluated; δ 技术 is the video AI and alarm analysis early warning coefficient for the target to be evaluated; α is a preset process risk weight coefficient, and β is a preset site risk weight coefficient; The intrinsic risk base value The determination formula is: ； wherein R 基准 is a preset reference value of the target to be evaluated; is a risk amplification factor of an i-th hazard source included in the target to be evaluated, is a state score of the i-th hazard source included in the target to be evaluated, the risk amplification factor being associated with a kind of the corresponding hazard source, and the state score being associated with an aging degree and a corrosion rate of the corresponding hazard source. The process production risk coefficient R 工艺 The determination formula is: ； wherein, is a real-time temperature monitoring value for the target to be evaluated; is a real-time pressure monitoring value for the target to be evaluated; is a real-time liquid level monitoring value for the target to be evaluated; is a preset temperature standard value for the target to be evaluated; is a preset pressure standard value for the target to be evaluated; is a preset liquid level standard value for the target to be evaluated; is a temperature allowable fluctuation range for the target to be evaluated; is a pressure allowable fluctuation range for the target to be evaluated; is a liquid level allowable fluctuation range for the target to be evaluated; The work site flammable / toxic gas risk coefficient R 场所 The determination formula is: ； ； wherein, a real-time concentration of combustible gas in the working site where the target to be evaluated is located, a real-time concentration of toxic gas in the working site where the target to be evaluated is located, LEL is the minimum critical value of the combustible gas reaching the explosive dangerous concentration in the working site where the target to be evaluated is located; IDLH is the minimum critical value of the toxic gas threatening the life and health concentration in the working site where the target to be evaluated is located. The management performance and hidden danger management correction coefficient γ 管理 The determination formula is: ； wherein, a package performance weight for the target to be evaluated, a hidden danger management weight for the target to be evaluated, the package performance weight being associated with a safety inspection frequency and an inspection record integrity for the target to be evaluated, and the hidden danger management weight being associated with an overdue unrectification duration and a total number of hidden dangers for the target to be evaluated; The video AI and the alarm analysis early warning coefficient δ 技术 The determination formula is: ； ； ； ； ； wherein the unit of the delay response time is min, and the total number of alarms is the sum of the effective early warning times and the unassociated alarm number.

Images