Multi-modal analysis-based major hazard source risk assessment method and system

Abstract

The invention provides a major hazard source risk assessment method and system based on multi-modal analysis, and the method comprises the steps: determining a certain major hazard source as a to-be-assessed target, obtaining the multi-modal monitoring data of the to-be-assessed target, and determining the comprehensive risk value of the to-be-assessed target according to the multi-modal monitoring data; therefore, the risk level of the to-be-assessed target is determined according to the comprehensive risk value of the to-be-assessed target, the risk level of the major hazard source can be rapidly and accurately determined, real-time assessment analysis and display of the security risk of the major hazard source are realized, and sending, checking, feedback and supervision of early warning information can be timely and automatically completed according to the early warning level. That is to say, the method provided by the invention can quickly and accurately realize risk assessment of major hazard sources, improve accuracy and timeliness, and meet risk assessment requirements of users in different usage scenarios.

Description

Technical Field

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

[0002] Current risk management and early warning technologies for major hazardous sources are gradually shifting toward intelligent and digital approaches. Existing technologies primarily utilize IoT sensor networks to collect key parameters such as temperature, pressure, and leak concentration in real time, combining them with big data analysis to construct risk assessment indicator systems (e.g., analytic hierarchy process and fuzzy comprehensive evaluation). Some systems also incorporate machine learning algorithms (e.g., LSTM and random forest) to mine features from historical accident data. Regarding early warning mechanisms, most solutions employ a multi-level threshold-triggered alarm model, and some advanced systems integrate GIS visualization platforms and emergency response databases. In recent years, the fusion of multi-source heterogeneous data (e.g., 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 defects still exist in related 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; moreover, the early warning models in related technologies have weak generalization capabilities, poor adaptability to complex scenarios (such as multiple hazardous source coupling, nonlinear equipment failure), high false alarm / missing alarm rates, and model training relies on historical data, making it difficult to cope with sudden abnormal working conditions; in addition, related technologies only independently evaluate a single hazardous source, and do not fully consider the dynamic correlation of multi-dimensional risks such as equipment, environment, and human operation, resulting in insufficient accuracy in global risk prediction; and lack autonomous learning and optimization capabilities, making it impossible to adjust risk weights according to real-time working conditions, and the early warning response strategy is rigid, making it difficult to adapt to long-term evolving risks such as production process changes or equipment aging. Therefore, there are problems with the accuracy and timeliness of risk assessment methods for major hazardous sources.

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

[0005] The present invention provides a major hazard source risk assessment method and system based on multimodal analysis, which can quickly and accurately perform risk assessment of major hazard sources, improve accuracy and timeliness, and meet users' risk assessment needs in different usage scenarios.

[0006] The technical solution of the present invention to solve the above technical problems is as follows: In the first aspect, the present invention provides a major hazard source risk assessment method based on multimodal analysis, including: obtaining monitoring data of the target to be assessed; the target to be assessed includes multiple hazard sources, and the hazard source includes at least one of a storage tank, a device, and a hazardous chemical warehouse, and the monitoring data includes the liquid level, temperature, pressure data of the corresponding hazard source, and real-time monitoring data and alarm data of the flammable and toxic gas concentration; according to the monitoring data of the target to be assessed, the process production process risk coefficient, the workplace flammable / toxic gas risk coefficient, the management performance and hidden danger control correction coefficient and the video AI and alarm analysis warning coefficient are determined, and the process production process risk coefficient is used to characterize the risk level of the hazard source in the process production process, and the workplace The flammable / toxic gas risk coefficient is used to characterize the risk level of flammable / toxic gases in the workplace of the hazardous source. The management performance and hidden danger control correction coefficient is used to characterize the impact of management performance and hidden danger control on the risk of the hazardous source. The video AI and alarm analysis and warning coefficient is used to characterize the risk level in the video data of the hazardous source. The comprehensive risk value of the target to be evaluated is determined based on the process production risk coefficient, the flammable / toxic gas risk coefficient of the workplace, the management performance and hidden danger control correction coefficient and the video AI and alarm analysis and warning coefficient. The risk level of the target to be evaluated is determined based on the comprehensive risk value of the target to be evaluated, where different risk levels correspond to different numerical ranges of the comprehensive risk value.

[0007] On the basis of the above technical solution, the present invention can also be improved as follows.

[0008] Furthermore, the risk levels include major risk, greater risk, general risk and low risk; the risk level of the target to be assessed is determined based on the comprehensive risk value of the target to be assessed, including: when the comprehensive risk value of the target to be assessed is greater than or equal to the first risk value, determining the risk level of the target to be assessed as major risk; when the comprehensive risk value of the target to be assessed is greater than or equal to the second risk value and less than the first risk value, determining the risk level of the target to be assessed as greater risk; when the comprehensive risk value of the target to be assessed is greater than or equal to the third risk value and less than the second risk value, determining the risk level of the target to be assessed as general risk; when the comprehensive risk value of the target to be assessed is less than the third risk value, determining the risk level of the target to be assessed as low risk.

[0009] Furthermore, the formula for determining the comprehensive risk value R is: ； Among them, R 固有 is the inherent risk base value of the target to be assessed; R 工艺 is the risk factor of the target to be evaluated during the process production; R 场所is the flammable / toxic gas risk factor of the workplace where the target to be assessed is located; γ 管理 is the correction coefficient for management performance and hidden danger control of the target to be evaluated; δ 技术 is the video AI and alarm analysis warning coefficient for the target to be evaluated; α and β are the preset process and site risk weight coefficients.

[0010] Furthermore, the inherent risk base value R 固有 The formula for determining is: ; Among them, R 基准 It is the preset benchmark value of the target to be evaluated; is the risk amplification factor of the ith hazard source included in the target to be assessed, The status score of the i-th hazard source included in the target to be evaluated is associated with the risk amplification factor and 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.

[0011] Process production risk factor R 工艺 The formula for determining is: ; in, is the real-time temperature monitoring value of the target to be evaluated; is the real-time pressure monitoring value for the target to be evaluated; is the real-time liquid level monitoring value of the target to be evaluated; is the preset temperature standard value for the target to be evaluated; is the preset pressure standard value for the target to be evaluated; is the preset liquid level standard value for the target to be evaluated; The permissible temperature fluctuation range for the target to be evaluated; The permissible fluctuation range for the pressure against the target to be assessed; The permissible fluctuation range of the liquid level for the target to be evaluated; Risk factor R of flammable / toxic gas in the workplace 场所 The formula for determining is: ; ; in, The real-time concentration of combustible gas in the workplace where the target to be evaluated is located. It is the real-time concentration of toxic gases in the workplace where the target to be evaluated is located. LEL is the minimum critical value of whether the combustible gas in the workplace where the target to be evaluated is located has reached the explosion hazard concentration; IDLH is the minimum critical value of the toxic gas concentration that threatens life and health in the workplace where the target to be evaluated is located.

[0012] Correction coefficient γ for management performance and hidden danger control 管理 The formula for determining is: ; in, To determine the weight of the performance of the guarantee for the target to be evaluated, The weight of hidden danger management for the target to be assessed, the weight of guarantee performance is related to the frequency of safety inspections and the completeness of inspection records for the target to be assessed, and the weight of hidden danger management is related to the duration of overdue rectification and the total number of hidden dangers for the target to be assessed; Video AI and alarm analysis warning coefficient δ 技术 The formula for determining is: ; ; ; ; ; The unit of delayed response time is min, and the total number of alarms is the sum of the number of valid warnings and the number of unassociated alarms.

[0013] 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, obtaining time series data of the real-time concentration of the target gas in the target area in the target time period, the target area being the area where any hazardous source included in the target to be evaluated is located; determining the change slope 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 in the target time period; determining the interference pattern corresponding to the change slope of the real-time concentration of the target gas in the target area from the short-time spike interference feature library; wherein the short-time spike interference feature library stores the interference pattern corresponding to each of the multiple change slopes; when the change slope of the real-time concentration of the target gas in the target area does not exist in the short-time spike interference feature library, the number of effective warnings is increased by 1; when the change slope of the real-time concentration of the target gas in the target area exists in the short-time spike interference feature library, the number of unassociated alarms is increased by 1.

[0014] Furthermore, when the slope of the change of the real-time target gas concentration of the target area does not exist in the short-time spike interference feature library, the number of effective warnings is increased by 1, including: when the slope of the change of the real-time target gas concentration of the target area does not exist in the short-time spike interference feature library, the real-time target gas concentrations of multiple areas associated with the target area are obtained; the change rate of the real-time target gas concentrations of the multiple areas is determined; when the number of areas where the change rate of the real-time target gas concentration is less than a preset change rate threshold is greater than or equal to a preset number threshold, the number of effective warnings is increased by 1; when the number of areas where the change rate of the real-time target gas concentration is less than the preset change rate threshold is less than the preset number threshold, the number of unassociated alarms is increased by 1.

[0015] 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, obtaining process parameter information, microclimate information and video data of the target area, the process parameter information includes reactor temperature, pressure and liquid level information, and the microclimate information includes wind speed and temperature and humidity information; determining the leakage probability of the target gas based on a pre-trained Bayesian neural network; when the leakage probability of the target gas is greater than or equal to the preset probability threshold, adding 1 to the number of effective warnings; when the leakage probability of the target gas is less than the preset probability threshold, adding 1 to the number of unassociated alarms.

[0016] Furthermore, the above method also includes: when the risk level is a major risk, starting the corresponding material allocation system according to the emergency response plan corresponding to the major risk level, sending a first alarm message to multiple terminal devices, the first alarm message is used to prompt that the risk level is a major risk, and establishing communication with multiple terminal devices to realize information transmission between multiple terminal devices; when the risk level is a large risk, general risk or low risk, sending a second alarm message to multiple terminal devices according to the emergency response plan corresponding to the large risk, general risk or low risk level, the second alarm message is used to prompt that the risk level is a large risk, general risk or low risk.

[0017] Furthermore, the above method also includes: when the risk level of the target to be assessed is a major risk, a first identifier is displayed on the preset interface, and the first identifier is red; when the risk level of the target to be assessed is a large risk, a second identifier is displayed on the preset interface, and the second identifier is orange; when the risk level of the target to be assessed is a general risk, a third identifier is displayed on the preset interface, and the third identifier is yellow; when the risk level of the target to be assessed is a low risk, a fourth identifier is displayed on the preset interface, and the fourth identifier is blue.

[0018] The beneficial effects of the present invention are: By acquiring multimodal monitoring data of the target to be assessed (major hazard source) and determining the comprehensive risk value of the target to be assessed based on the multimodal monitoring data; and then determining the risk level of the target to be assessed based on the comprehensive risk value of the target to be assessed, the risk level of the major hazard source can be quickly and accurately determined, and real-time assessment, analysis, and display of the safety risks of the major hazard source can be achieved. It also supports the automatic and immediate sending, verification, feedback, and supervision of warning information based on the warning level. In other words, the method provided by the present invention can quickly and accurately perform risk assessment of major hazard sources, improve accuracy and timeliness, and meet the risk assessment needs of users in different usage scenarios.

[0019] In a second aspect, the present invention provides a major hazard source risk assessment system based on multimodal analysis, comprising: A data acquisition module is used to acquire monitoring data of the target to be evaluated; the target to be evaluated includes multiple hazardous sources, including at least one of a storage tank, a device, and a hazardous chemical warehouse. The monitoring data includes real-time monitoring data and alarm data of the liquid level, temperature, pressure data, and combustible and toxic gas concentration of the corresponding hazardous source; The risk value determination module is used to dynamically correct the process production risk coefficient, the flammable / toxic gas risk coefficient of the workplace, the management performance and hidden danger control correction coefficient, and the video AI and alarm analysis warning coefficient according to the fluctuation of the monitoring data of the target to be evaluated. The process production risk coefficient is used to characterize the risk level in the process production process of the hazardous source, the flammable / toxic gas risk coefficient in the workplace is used to characterize the risk level of flammable / toxic gases in the workplace of the hazardous source, the management performance and hidden danger control correction coefficient is used to characterize the impact of management performance and hidden danger control on the risk of the hazardous source, and the video AI and alarm analysis warning coefficient is used to characterize the risk level in the video data of the hazardous source.

[0020] 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 factor, the flammable / toxic gas risk factor of the workplace, the management performance and hidden danger control correction factor, and the video AI and alarm analysis warning factor.

[0021] The risk level determination module is used to determine the risk level of the target to be assessed according to the comprehensive risk value of the target to be assessed, wherein different risk levels correspond to different numerical ranges of the comprehensive risk value.

[0022] In a third aspect, the present invention provides an electronic device comprising: a memory, one or more processors; the memory and the processor are coupled; wherein the memory stores computer program code, the computer program code comprises computer instructions, and when the computer instructions are executed by the processor, the electronic device executes the personnel risk assessment method of any one of the above-mentioned first aspects.

[0023] In a fourth aspect, a computer-readable storage medium is provided, comprising computer instructions. When the computer instructions are executed on an electronic device, the electronic device executes the personnel risk assessment method of any one of the above-mentioned first aspects.

[0024] In a fifth aspect, a computer program product is provided. When the computer program product is run on a computer, the computer is caused to execute any one of the personnel risk assessment methods of the first aspect.

[0025] It can be understood that the beneficial effects that can be achieved by the personnel risk assessment system of the second aspect, the electronic device 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 possible design method thereof, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A schematic diagram of a process flow of a major hazard source risk assessment method based on multimodal analysis provided by the present invention; Figure 2 A schematic diagram of a process flow of another major hazard source risk assessment method based on multimodal analysis provided by the present invention; Figure 3 This is a schematic diagram of a major hazard source risk assessment system provided by the present invention. DETAILED DESCRIPTION

[0027] The following describes the technical solutions in the embodiments of the present application in conjunction with the accompanying drawings. In the description of this application, unless otherwise specified, " / " indicates that the objects associated with each other are in an "or" relationship. For example, A / B can mean A or B. "And / or" in this application is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, "A and / or B" can mean: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, in the description of this application, unless otherwise specified, "multiple" means two or more. "At least one of the following" or similar expressions refers 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 mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural. In addition, in order to facilitate the clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, words such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. Those skilled in the art will understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit differences. At the same time, in the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or design.

[0028] Current risk management and early warning technologies for major hazardous sources are gradually shifting toward intelligent and digital approaches. Existing technologies primarily utilize IoT sensor networks to collect key parameters such as temperature, pressure, and leak concentration in real time, combining them with big data analysis to construct risk assessment indicator systems (e.g., analytic hierarchy process and fuzzy comprehensive evaluation). Some systems also incorporate machine learning algorithms (e.g., LSTM and random forest) to mine features from historical accident data. Regarding early warning mechanisms, most solutions employ a multi-level threshold-triggered alarm model, and some advanced systems integrate GIS visualization platforms and emergency response databases. In recent years, the fusion of multi-source heterogeneous data (e.g., equipment status data, meteorological and environmental data, and video surveillance data) and the optimization of dynamic risk assessment models have become research hotspots.

[0029] However, the following defects still exist in related 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; moreover, the early warning models in related technologies have weak generalization capabilities, poor adaptability to complex scenarios (such as multiple hazardous source coupling, nonlinear equipment failure), high false alarm / missing alarm rates, and model training relies on historical data, making it difficult to cope with sudden abnormal working conditions; in addition, related technologies only independently evaluate a single hazardous source, and do not fully consider the dynamic correlation of multi-dimensional risks such as equipment, environment, and human operation, resulting in insufficient accuracy in global risk prediction; and lack autonomous learning and optimization capabilities, making it impossible to adjust risk weights according to real-time working conditions, and the early warning response strategy is rigid, making it difficult to adapt to long-term evolving risks such as production process changes or equipment aging. Therefore, there are problems with the accuracy and timeliness of risk assessment methods for major hazardous sources.

[0030] Therefore, there is an urgent need for a major hazard source risk assessment method and system based on multimodal analysis, which can quickly and accurately perform risk assessment of major hazard sources, improve accuracy and timeliness, and meet users' risk assessment needs in different usage scenarios.

[0031] For the above issues, see Figure 1 The present invention provides a major hazard source risk assessment method based on multimodal analysis, comprising steps S101-S104: S101: Acquire monitoring data of the target to be evaluated.

[0032] Specifically, the targets to be evaluated include multiple hazardous sources, including at least one of storage tanks, equipment, and hazardous chemical warehouses. The monitoring data include real-time monitoring data and alarm data of the liquid level, temperature, pressure data, and combustible and toxic gas concentrations of the corresponding hazardous sources.

[0033] It should be understood that the target to be evaluated is a major hazard source including multiple hazard sources. Multiple hazard sources can be divided into one major hazard source based on the setting area, multiple hazard sources can also be divided into one major hazard source based on different parts of the process flow, and multiple hazard sources can also be divided into one major hazard source based on the projects to which they belong. The embodiment of the present invention does not impose any special restrictions on the specific division method of major hazard sources.

[0034] In one example, the specific implementation method for acquiring monitoring data for the target to be assessed involves collaborative multimodal data acquisition, integrating heterogeneous sensor data (including level, pressure, and temperature sensors, and toxic and combustible gas concentration detectors) and video surveillance information. Edge computing nodes are then used for data normalization and preprocessing to eliminate data heterogeneity and latency differences. Alternatively, this monitoring data can be understood as sensor monitoring data from major hazardous sources (targets to be assessed), including real-time monitoring data on liquid level, pressure, temperature, combustible gas detection, and toxic gas detection, as well as alarm data for liquid level, temperature, pressure, and toxic and combustible gas concentrations in storage tanks, equipment, and hazardous chemical storage facilities. Historical data can be queried and compared to provide data support for online spot checks of major hazardous sources. Alarm data, including video surveillance footage from key locations such as ammonium nitrate warehouses, central control rooms, and major hazardous source sites, provides intelligent analysis of surveillance video. This provides data support for comprehensive identification and early warning of fires, smoke, and personnel violations (e.g., absent personnel from the central control room).

[0035] In one possible implementation, the method provided by the embodiment of the present invention further includes: The monitoring data of the evaluation targets shall be cleaned and sorted to ensure the integrity and consistency of the data.

[0036] Specifically, it integrates monitoring data from major hazardous sources, including basic information on major hazardous sources, AI information from surveillance videos, sensor monitoring data, data on performance of security duties, and data on hidden danger investigation and management, and presents the results to end users. It has cross-cloud, cross-network, and cross-data center data integration capabilities, provides a programming interface / gateway for applications, supports message publishing and subscription, multi-cluster deployment, and message trajectory tracking. Based on data integration, it conducts a variety of data mining and analysis, including descriptive analysis, diagnostic analysis, predictive analysis, and causal analysis; it provides a variety of analytical methods, models, and tools, including statistical analysis, retrieval, machine learning, text analysis, and video analysis, to help optimize the construction and management of the Industrial Internet Intelligent Supervision Platform.

[0037] It should be noted that basic information on major hazardous sources refers to the essential data and information used to describe and identify them. This information is crucial for their assessment, monitoring, and management. This information typically includes the following aspects: a basic description of the source (including its name and location, the type, quantity, and characteristics of the hazardous substances involved, and the scale and methods of its production, storage, use, or operation); the types of accidents that could occur at the source; the inherent risk level and potential impact of the source; emergency rescue and accident handling plans and measures; and relevant laws, regulations, and standards for hazard management.

[0038] Major hazard source surveillance video AI uses artificial intelligence (AI) to analyze and process video data from major hazard sites in real time, enabling intelligent monitoring, early warning, and response. This technology, which combines computer vision, machine learning, deep learning, and other AI techniques, can automatically identify and analyze anomalies in video, such as smoke, fire, and personnel sleeping or leaving their posts, improving monitoring accuracy and efficiency.

[0039] Major hazard source sensor monitoring refers to the use of various types of sensors to monitor key parameters at major hazard source sites in real time. According to monitoring needs, appropriate sensor types are selected, such as temperature sensors, pressure sensors, liquid level sensors, gas sensors (to detect combustible gases and toxic gases), vibration sensors, flow sensors, etc.

[0040] Sensor monitoring is an important part of the safety management of major hazardous sources. It can provide real-time and accurate data to help managers understand the status of hazardous sources and take necessary preventive measures.

[0041] The responsibility for ensuring the safety of major hazardous sources means that in order to ensure the safety of major hazardous sources, key personnel such as the company's main person in charge, technical person in charge, and operation person in charge will conduct all-round and full-process supervision and management of major hazardous sources to ensure that the safety risks of major hazardous sources are effectively controlled and accidents are prevented.

[0042] The responsibilities for guaranteeing major hazardous sources include the following: Establish and improve a safety guarantee responsibility system for major hazardous sources, clarifying the safety responsibilities of responsible individuals at all levels. Develop and improve safety management systems and operating procedures for major hazardous sources to ensure their effective implementation. Regularly conduct safety inspections and assessments of major hazardous sources to promptly identify and eliminate potential safety hazards. Develop and implement emergency response plans for major hazardous sources to ensure timely and effective disposal and rescue efforts in the event of an accident. Strengthen safety training and education for major hazardous sources to enhance employee safety awareness and operational skills. Regularly organize drills for major hazardous sources to test the effectiveness of emergency response plans and employee emergency response capabilities. Regularly summarize and evaluate the safety management of major hazardous sources to continuously improve and enhance safety management effectiveness.

[0043] The investigation and control of hidden dangers of major hazardous sources refers to a comprehensive and systematic inspection of major hazardous sources, the identification of existing safety hazards, and the adoption of effective measures to rectify them in order to eliminate or reduce safety hazards and ensure the safe operation of major hazardous sources.

[0044] The investigation and control of major hazardous sources and hidden dangers include the following: Develop a Hazard Inspection Plan: Based on the characteristics, scale, and potential risks of major hazardous sources, a detailed hazard inspection plan should be developed, clearly defining the scope, content, methods, and timeline of the inspection. Conduct Hazard Inspections: In accordance with the hazard inspection plan, professional personnel will conduct a comprehensive and systematic inspection of major hazardous sources, covering equipment and facilities, process flows, safety precautions, and emergency rescue facilities. Identify Safety Hazards: During the hazard inspection process, through observation, inspection, and testing, existing safety hazards will be identified, including defects in equipment and facilities, unreasonable process flows, and inadequate safety precautions. Analyze Safety Hazards: Conduct an in-depth analysis of identified safety hazards to determine their causes, scope, and severity, and assess their risks to the safe operation of the major hazardous source. Develop a Corrective Action Plan: Based on the results of the safety hazard analysis, a corresponding corrective action plan will be developed, clearly defining the objectives, measures, responsibilities, and timelines for the corrective actions. Implement Hazard Correction: In accordance with the corrective action plan, personnel will be organized to rectify the safety hazards, including repairing equipment and facilities, optimizing process flows, and improving safety precautions. Verify the effectiveness of rectification: Verify the major hazard sources after rectification to ensure that safety hazards are effectively eliminated or reduced and meet the requirements of safe operation. Establish a hazard investigation and treatment file: Record and archive the process, results and rectification status of hazard investigation and treatment as an important basis for the safety management of major hazard sources.

[0045] S102: Determine the process production risk factor, the flammable / toxic gas risk factor in the workplace, the management performance and hidden danger control correction factor, and the video AI and alarm analysis warning factor based on the monitoring data of the target to be evaluated.

[0046] Specifically, the process production risk coefficient is used to characterize the risk level in the process production of hazardous sources; the workplace flammable / toxic gas risk coefficient is used to characterize the risk level of flammable / toxic gases in the workplace of hazardous sources; the management performance and hidden danger control correction coefficient is used to characterize the impact of management performance and hidden danger control on the risk of hazardous sources; the video AI and alarm analysis warning coefficient is used to characterize the risk level in the video data of hazardous sources; Process production risk factor R 工艺 The formula for determining is: ; in, is the real-time temperature monitoring value; is the real-time pressure monitoring value; is the real-time liquid level monitoring value; It is the preset temperature standard value; It is the preset pressure standard value; It is the preset liquid level standard value; The allowable temperature fluctuation range; The allowable pressure fluctuation range; The allowable fluctuation range of the liquid level; In one example, .

[0047] It should be noted that when the threshold is exceeded, R 工艺 Counted as 100%, dynamic correction is triggered.

[0048] Risk factor R of flammable / toxic gas in the workplace 场所 The formula for determining is: ; ; in, is the real-time concentration of combustible gas, It is the real-time concentration of toxic gases. LEL is the minimum critical value for whether the combustible gas reaches the explosion hazard concentration; IDLH is the minimum critical value for the concentration of toxic gases that threaten life and health.

[0049] It's important to note that the LEL (Less Elevated Light Emission Limit) is the lowest concentration of a flammable gas or vapor in air. When this concentration is reached or exceeded, it could cause an explosion or combustion upon contact with an ignition source (spark, high temperature, etc.). It's typically expressed as a percentage by volume (%vol). For example, the LEL for methane is 5% (meaning an explosion risk exists when the methane concentration in air is ≥5%). The LEL is the minimum critical value for determining whether a flammable gas has reached a dangerously explosive concentration.

[0050] IDLH stands for Immediately Dangerous to Life and Health. It is the concentration level of a toxic substance in the air at which short-term exposure (usually ≤30 minutes) could cause irreversible health damage, loss of ability to escape, or death. It is expressed in ppm (parts per million) or mg / m³. For example, the IDLH for chlorine is 10 ppm. The IDLH is an important metric for determining whether to wear emergency escape respirators.

[0051] Correction coefficient γ for management performance and hidden danger control 管理 The formula for determining is: ; in, To ensure the weight of performance of duties, The weight of hidden danger management is related to the frequency of safety inspections and the completeness of inspection records. The weight of hidden danger management is related to the length of time that no rectification has been carried out within the prescribed period and the total number of hidden dangers. Specifically, .

[0052] Video AI and alarm analysis warning coefficient δ 技术 The formula for determining is: ; ; ; ; ; The unit of delayed response time is min, and the total number of alarms is the sum of the number of valid warnings and the number of unassociated alarms.

[0053] The method provided by the embodiment of the present invention dynamically couples multiple coefficients. The multiple influences of equipment operation, human factors and technology solve the problems of insufficient data integration and real-time performance of traditional algorithms, lack of risk coupling analysis and low level of adaptability and intelligence. In addition, if any indicator (such as toxic gas) exceeds the limit, the warning level can be directly upgraded. At the same time, the associated impact of the warning is also considered, making the risk assessment more diverse and reasonable, and solving the problem of weak generalization ability of the warning model of the traditional algorithm. Finally, the method provided by the embodiment of the present invention supports adjusting the weights , which can be adapted to different industry scenarios and meet users' usage needs in different usage scenarios.

[0054] S103: Determine the comprehensive risk value of the target to be assessed based on the process risk factor, the flammable / toxic gas risk factor in the workplace, the management performance and hidden danger control correction factor, and the video AI and alarm analysis warning factor; In some embodiments, the formula for determining the comprehensive risk value R is: ； Among them, R 固有 is the inherent risk base value of the target to be assessed; R 工艺 is the risk factor of the target to be evaluated during the process production; R 场所 is the flammable / toxic gas risk factor of the workplace where the target to be assessed is located; γ 管理 is the correction coefficient for management performance and hidden danger control of the target to be evaluated; δ 技术 is the video AI and alarm analysis warning coefficient for the target to be evaluated; α and β are the preset process and site risk weight coefficients.

[0055] In one example, α=0.6 and β=0.4.

[0056] Furthermore, the inherent risk base value R 固有 The formula for determining is: ; Among them, R 基准 It is the preset benchmark value of the target to be evaluated; is the risk amplification factor of the ith hazard source included in the target to be assessed, The status score of the i-th hazard source included in the target to be evaluated is associated with the risk amplification factor and 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.

[0057] In one example, for hazardous toxic substances, k1=0.2, and for hazardous high-voltage equipment, k2=0.15.

[0058] In some embodiments, the method provided by the embodiment of the present invention further includes: When the real-time concentration of the target gas in the target area is greater than or equal to a preset threshold, the time series data of the real-time concentration of the target gas in the target area in the target time period is obtained, and the target area is the area where any hazardous source included in the target to be evaluated is located; the change slope of the real-time concentration of the target gas in the target area is determined according to the time series data of the real-time concentration of the target gas in the target area in the target time period; the interference pattern corresponding to the change slope of the real-time concentration of the target gas in the target area is determined from the short-time spike interference feature library; wherein the short-time spike interference feature library stores the interference pattern corresponding to each change slope of a plurality of change slopes; when the change slope of the real-time concentration of the target gas in the target area does not exist in the short-time spike interference feature library, the number of effective warnings is increased by 1; when the change slope of the real-time concentration of the target gas in the target area exists in the short-time spike interference feature library, the number of unassociated alarms is increased by 1.

[0059] Furthermore, when the slope of the change of the real-time concentration of the target gas in the target area does not exist in the short-time spike interference feature library, the number of valid warnings is increased by 1, including: When the slope of the change of the real-time target gas concentration of the target area does not exist in the short-time spike interference feature library, the real-time target gas concentrations of multiple areas associated with the target area are obtained; the change rate of the real-time target gas concentrations of the multiple areas is determined; when the number of areas where the change rate of the real-time target gas concentration is less than a preset change rate threshold is greater than or equal to a preset number threshold, the number of effective warnings is increased by 1; when the number of areas where the change rate of the real-time target gas concentration is less than the preset change rate threshold is less than the preset number threshold, the number of unassociated alarms is increased by 1.

[0060] In some other embodiments, the method provided by the embodiment of the present invention further includes: When the real-time concentration of the target gas in the target area is greater than or equal to the preset threshold, the process parameter information, microclimate information and video data of the target area are obtained. The process parameter information includes the reactor temperature, pressure and liquid level information, and the microclimate information includes the wind speed and temperature and humidity information. The leakage probability of the target gas is determined based on the pre-trained Bayesian neural network. When the leakage probability of the target gas is greater than or equal to the preset probability threshold, the number of effective warnings is increased by 1. When the leakage probability of the target gas is less than the preset probability threshold, the number of unassociated alarms is increased by 1.

[0061] As can be seen from the above, the method provided by the embodiment of the present invention addresses the instantaneous false alarm problem existing in the monitoring of toxic and combustible gases in major hazardous source areas and constructs a multi-dimensional intelligent identification system: two filtering mechanisms are used to determine false alarms: First, based on the primary screening of time series pattern recognition, the sliding time window algorithm is used to analyze the deviation between the mutation slope and duration of the concentration value and the GB / T 50493 standard threshold, and a short-term spike interference feature library is established. Based on the short-term spike interference feature library, an accurate judgment is made on whether it is a valid alarm. In addition, when a single-point alarm (target area) is triggered, the concentration change trend of the adjacent monitoring units (within a radius of 50 meters) is automatically checked. If no spatial correlation gradient is formed, a false alarm mark is triggered; alternatively, multi-system data coupling analysis is sampled, and the enterprise DCS system is linked to obtain process parameters (such as reactor temperature, pressure, liquid level, etc.), meteorological station microclimate data (wind speed, temperature and humidity) and video surveillance intelligent recognition results. A causal inference model is constructed using a Bayesian network to calculate the true leakage probability value.

[0062] After actual measurement and verification, this multi-dimensional intelligent identification system reduces the false alarm rate by more than 90%. At the same time, a self-learning optimization module is established to extract features of historical false alarm cases through the platform, and continuously update the dynamic compensation coefficient of equipment sensitivity and the environmental interference factor weight library.

[0063] S104: Determine the risk level of the target to be assessed according to the comprehensive risk value of the target to be assessed, wherein different risk levels correspond to different numerical ranges of the comprehensive risk value.

[0064] In some embodiments, see Figure 2 The risk levels include major risk, relatively high risk, general risk and low risk; the above S104 includes: S1041: When the comprehensive risk value of the target to be assessed is greater than or equal to the first risk value (for example, 61), the risk level of the target to be assessed is determined to be a significant risk; S1042: When the comprehensive risk value of the target to be assessed is greater than or equal to the second risk value (for example, 42) and less than the first risk value (for example, 61), the risk level of the target to be assessed is determined to be a high risk. S1043: When the comprehensive risk value of the target to be assessed is greater than or equal to the third risk value (for example, 21) and less than the second risk value (for example, 61), the risk level of the target to be assessed is determined to be general risk; S1044: When the comprehensive risk value of the target to be evaluated is less than the third risk value (for example, 21), the risk level of the target to be evaluated is determined to be low risk.

[0065] In some embodiments, the present invention further includes: In the case where the risk level is a major risk, the corresponding material allocation system is activated according to the emergency response plan corresponding to the major risk level, a first alarm message is sent to multiple terminal devices, the first alarm message is used to prompt that the risk level is a major risk, and communication is established with the multiple terminal devices to realize information transmission between the multiple terminal devices; When the risk level is a high risk, a general risk or a low risk, a second alarm message is sent to multiple terminal devices according to the emergency response plan corresponding to the high risk, the general risk or the low risk. The second alarm message is used to prompt that the risk level is a high risk, a general risk or a low risk.

[0066] It can also be understood that: when the risk level is major risk, the method provided by the present invention will, after completing the precise matching of the emergency plan, simultaneously start the material allocation system, communication guarantee module and expert consultation mechanism associated with the plan, forming a full-process closed-loop management of "monitoring and early warning-intelligent matching-solution generation-collaborative disposal", significantly improving the timeliness of emergency response and the scientific nature of the disposal plan.

[0067] In addition, the method provided by the embodiment of the present invention requires joint handling when the risk level is low risk, general risk, or high risk. The platform will automatically notify the relevant responsible persons of the enterprise, including the security manager, safety officer, etc.

[0068] In some embodiments, the present invention further includes: When the risk level is a major risk, the first mark is displayed on the preset interface, and the first mark is red; when the risk level is a large risk, the second mark is displayed on the preset interface, and the second mark is orange; when the risk level is a general risk, the third mark is displayed on the preset interface, and the third mark is yellow; when the risk level is a low risk, the fourth mark is displayed on the preset interface, and the fourth mark is blue.

[0069] As can be seen from the above S101-S104, the method provided by the embodiment of the present invention obtains multimodal monitoring data of the target to be evaluated, and determines the comprehensive risk value of the target to be evaluated based on the multimodal monitoring data; and then determines the risk level of the target to be evaluated based on the comprehensive risk value of the target to be evaluated, which can quickly and accurately determine the risk level of major hazardous sources, realize real-time assessment, analysis and display of safety risks of major hazardous sources, and support the automatic and immediate completion of the sending, verification, feedback and supervision of early warning information according to the early warning level. In other words, the method provided by the present invention can quickly and accurately realize the risk assessment of major hazardous sources, improve accuracy and timeliness, and meet the risk assessment needs of users in different usage scenarios.

[0070] See also Figure 3 The present invention also provides a major hazard source risk assessment system based on multimodal analysis, comprising: a data acquisition module for acquiring monitoring data of a target to be assessed; the target to be assessed includes multiple hazard sources, and the hazard sources include storage tanks, devices, or hazardous chemical warehouses; the monitoring data includes real-time monitoring data and alarm data of the liquid level, temperature, pressure data, and combustible and toxic gas concentration of the corresponding hazard sources; The risk value determination module is used to determine the process production risk coefficient, the workplace flammable / toxic gas risk coefficient, the management performance and hidden danger control correction coefficient, and the video AI and alarm analysis warning coefficient based on the monitoring data of the target to be evaluated. The process production risk coefficient is used to characterize the risk level in the process production process of the hazardous source, the workplace flammable / toxic gas risk coefficient is used to characterize the risk level of flammable / toxic gases in the workplace of the hazardous source, the management performance and hidden danger control correction coefficient is used to characterize the impact of management performance and hidden danger control on the risk of the hazardous source, and the video AI and alarm analysis warning coefficient is used to characterize the risk level in the video data of the hazardous 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, the workplace flammable / toxic gas risk coefficient, the management performance and hidden danger control correction coefficient, and the video AI and alarm analysis warning coefficient; The risk level determination module is used to determine the risk level of the target to be assessed according to the comprehensive risk value of the target to be assessed, wherein different risk levels correspond to different numerical ranges of the comprehensive risk value.

[0071] In some schemes, multiple embodiments of the present application can be combined and the combined scheme can be implemented. Optionally, some operations in the process of each method embodiment are optionally combined, and / or the order of some operations is optionally changed. In addition, the execution order between the steps of each process is only exemplary and does not constitute a limitation on the execution order between the steps. There can also be other execution orders between the steps. It is not intended to indicate that the execution order is the only order in which these operations can be performed. Ordinary technicians in this field will think of many ways to reorder the operations of this article. In addition, it should be noted that the process details involved in a certain embodiment of this article are also applicable to other embodiments in a similar manner, or different embodiments can be used in combination.

[0072] Furthermore, some steps in the method embodiments may be equivalently replaced with other possible steps. Alternatively, some steps in the method embodiments may be optional and may be deleted in certain usage scenarios. Alternatively, other possible steps may be added to the method embodiments. Furthermore, the various method embodiments may be implemented separately or in combination.

[0073] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned 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.

[0074] In the several 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 example, the division of modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or units, which can be electrical, mechanical or other forms.

[0075] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0076] If the integrated unit is implemented in the form of 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 solution of the embodiment of the present application, or the part that makes the contribution, or all or part of the technical solution can be embodied in the form of a software product. The software product is stored in a storage medium and includes several instructions for enabling a device (which can be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0077] The above content is only a specific embodiment of this application, but the scope of protection of this application is not limited to this. Any changes or replacements within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A risk assessment method for major hazardous sources based on multimodal analysis, characterized in that: include: Acquire monitoring data of a target to be assessed; the target to be assessed includes multiple hazardous sources, including at least one of a storage tank, a device, and a hazardous chemical warehouse; the monitoring data includes real-time monitoring data and alarm data of liquid level, temperature, pressure data, and combustible and toxic gas concentrations of the corresponding hazardous source; Based on the monitoring data of the target to be evaluated, determine the process production process risk coefficient, the workplace flammable / toxic gas risk coefficient, the management performance and hidden danger control correction coefficient, and the video AI and alarm analysis warning coefficient; wherein, the process production process risk coefficient is used to characterize the risk level in the process production process of the hazardous source, the workplace flammable / toxic gas risk coefficient is used to characterize the risk level of flammable / toxic gases in the workplace of the hazardous source, the management performance and hidden danger control correction coefficient is used to characterize the impact of management performance and hidden danger control on the risk of the hazardous source, and the video AI and alarm analysis warning coefficient is used to characterize the risk level in the video data of the hazardous source; Determine the comprehensive risk value of the target to be assessed based on the process risk factor, the flammable / toxic gas risk factor of the workplace, the management performance and hidden danger control correction factor, and the video AI and alarm analysis warning factor; The risk level of the target to be assessed is determined according to the comprehensive risk value of the target to be assessed, wherein different risk levels correspond to different numerical ranges of the comprehensive risk value.

2. The method according to claim 1, characterized in that The risk levels include major risk, relatively high risk, average risk and low risk; Determining the risk level of the target to be assessed based on the comprehensive risk value of the target to be assessed includes: If the comprehensive risk value of the target to be assessed is greater than or equal to the first risk value, determining the risk level of the target to be assessed as a major risk; If the comprehensive risk value of the target to be assessed is greater than or equal to the second risk value and less than the first risk value, determining the risk level of the target to be assessed as a high risk; If the comprehensive risk value of the target to be assessed is greater than or equal to the third risk value and less than the second risk value, the risk level of the target to be assessed is determined to be general risk; When the comprehensive risk value of the target to be assessed is less than the third risk value, the risk level of the target to be assessed is determined to be low risk.

3. The method according to claim 2, characterized in that The formula for determining the comprehensive risk value R is: ； Among them, R 固有 is the inherent risk base value of the target to be assessed; R 工艺 is the risk factor of the target to be evaluated in the process of production; R 场所 is the flammable / toxic gas risk factor of the workplace where the target to be assessed is located; γ 管理 is the correction coefficient for management performance and hidden danger control of the target to be evaluated; δ 技术 is the video AI and alarm analysis warning coefficient for the target to be evaluated; α is the preset process risk weight coefficient, and β is the preset site risk weight coefficient.

4. The method according to claim 3, characterized in that The inherent risk base value R 固有 The formula for determining is: ； Among them, R 基准 is a preset benchmark value of the target to be evaluated; is the risk amplification factor of the ith hazard source included in the target to be assessed, A status score for the i-th hazard source included in the target to be assessed, wherein 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; The risk factor R of the process 工艺 The formula for determining is: ； in, is the real-time temperature monitoring value of the target to be evaluated; is a real-time pressure monitoring value for the target to be evaluated; is the real-time liquid level monitoring value of 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; The allowable temperature fluctuation range for the target to be evaluated; The allowable fluctuation range of pressure for the target to be evaluated; is the allowable fluctuation range of the liquid level of the target to be evaluated; The flammable / toxic gas risk factor R of the workplace 场所 The formula for determining is: ； ； in, is the real-time concentration of combustible gas at the workplace where the target to be evaluated is located, The real-time concentration of toxic gases in the workplace where the target to be assessed is located. LEL is the lowest critical value of the combustible gas concentration at the workplace where the target to be assessed is located, which reaches an explosion hazard. IDLH is the lowest critical value of the toxic gas concentration at the workplace where the target to be assessed is located, which threatens life and health. The correction coefficient γ for management performance and hidden danger control 管理 The formula for determining is: ； in, The weight of the guarantee performance for the target to be evaluated, The weight of hidden danger management for the target to be assessed is related to the frequency of safety inspections and the completeness of inspection records for the target to be assessed, and the weight of hidden danger management is related to the duration of overdue rectification and the total number of hidden dangers for the target to be assessed; The video AI and alarm analysis warning coefficient δ 技术 The formula for determining is: ； ； ； ； ； The unit of delayed response time is min, and the total number of alarms is the sum of the number of valid warnings and the number of unassociated alarms.

5. The method according to claim 4, characterized in that The method further comprises: When the real-time concentration of the target gas in the target area is greater than or equal to a preset threshold, 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 included 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 time series data of the real-time concentration of the target gas in the target area in a target time period; Determining an interference pattern corresponding to a change slope of the real-time concentration of the target gas in the target area from a short-time spike interference feature library; wherein the short-time spike interference feature library stores an interference pattern corresponding to each of a plurality of change slopes; If the change slope of the real-time concentration of the target gas in the target area does not exist in the short-time spike interference feature library, the number of valid warnings is increased by 1; When the short-time spike interference feature library contains the change slope of the real-time concentration of the target gas in the target area, the number of unassociated alarms is increased by 1.

6. The method according to claim 5, characterized in that When the change slope of the real-time concentration of the target gas in the target area does not exist in the short-time spike interference feature library, the number of valid warnings is increased by 1, including: When the short-time spike interference feature library does not contain a change slope of the target gas real-time concentration of the target area, obtaining the target gas real-time concentrations of a plurality of areas associated with the target area; determining a rate of change of the real-time concentration of the target gas in the plurality of regions; When the number of regions where the change rate of the target gas real-time concentration is less than the preset change rate threshold is greater than or equal to the preset number threshold, the number of valid warnings is increased by 1; When the number of regions where the change rate of the real-time concentration of the target gas is less than the preset change rate threshold is less than the preset number threshold, the number of unassociated alarms is increased by 1.

7. The method according to claim 6, characterized in that The method further comprises: When the real-time concentration of the target gas in the target area is greater than or equal to a preset threshold, obtaining process parameter information, microclimate information, and video data of the target area, wherein the process parameter information includes reactor temperature, pressure, and liquid level information, and the microclimate information includes wind speed and temperature and humidity information; Determine the leakage probability of the target gas based on a pre-trained Bayesian neural network; When the leakage probability of the target gas is greater than or equal to the preset probability threshold, the number of valid warning times is increased by 1; When the leakage probability of the target gas is less than a preset probability threshold, the number of unassociated alarms is increased by 1.

8. The method according to claim 7, characterized in that The method further comprises: When the risk level is a major risk, initiating a corresponding material allocation system according to an emergency response plan corresponding to the major risk level, sending a first alarm message to a plurality of terminal devices, wherein the first alarm message is used to indicate that the risk level is a major risk, and establishing communication with the plurality of terminal devices to enable information transmission between the plurality of terminal devices; When the risk level is a greater risk, a general risk or a low risk, a second alarm message is sent to multiple terminal devices according to the emergency response plan corresponding to the greater risk, the general risk or the low risk, and the second alarm message is used to prompt that the risk level is a greater risk, a general risk or a low risk.

9. The method according to claim 8, characterized in that The method further comprises: If the risk level of the target to be assessed is a major risk, a first indicator is displayed on the preset interface, and the first indicator is red; If the risk level of the target to be assessed is a 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 a general 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.

10. A major hazard source risk assessment system based on multimodal analysis, characterized in that: include: A data acquisition module is configured to acquire monitoring data of a target to be evaluated; the target to be evaluated includes multiple hazardous sources, including at least one of a storage tank, a device, and a hazardous chemical warehouse; the monitoring data includes real-time monitoring data and alarm data of the liquid level, temperature, pressure, and combustible and toxic gas concentration of the corresponding hazardous source; A risk value determination module is used to determine, based on the monitoring data of the target to be evaluated, a process production process risk coefficient, a workplace flammable / toxic gas risk coefficient, a management performance and hidden danger control correction coefficient, and a video AI and alarm analysis warning coefficient; wherein the process production process risk coefficient is used to characterize the degree of risk in the process production process of the hazardous source, the workplace flammable / toxic gas risk coefficient is used to characterize the degree of risk of flammable / toxic gases in the workplace of the hazardous source, the management performance and hidden danger control correction coefficient is used to characterize the degree of influence of management performance and hidden danger control on the risk of the hazardous source, and the video AI and alarm analysis warning coefficient is used to characterize the degree of risk in the video data of the hazardous source; The risk value determination module is further configured to determine the comprehensive risk value of the target to be assessed based on the process risk factor, the flammable / toxic gas risk factor of the workplace, the management performance and hidden danger control correction factor, and the video AI and alarm analysis warning factor; The risk level determination module is used to determine the risk level of the target to be assessed according to the comprehensive risk value of the target to be assessed, wherein different risk levels correspond to different numerical ranges of the comprehensive risk value.

Images