Multi-source fusion industrial safety risk management and control and intelligent interaction method and system

Abstract

The invention relates to the field of industrial safety management and control, and discloses a multi-source fusion industrial safety risk management and control and intelligent interaction method and system. The method comprises the following steps: fusing multi-source data such as plant equipment operation parameters and environment monitoring data, and constructing an industrial scene dynamic risk level distribution diagram; obtaining the position of a target person, extracting the real-time risk information of an area where the target person is located and a planned path, and generating a personalized safety guidance sequence; generating a dynamic navigation path containing a risk prompt and an avoidance action through superposition of an augmented reality technology and guidance; if the risk of the path node exceeds the standard, recalculating a low-risk alternative navigation sequence; and extracting the updated risk information correction distribution map and adjusting a guidance model rule to form refined safety management data. According to the method, the problems of inaccurate guidance, high path risk and difficulty in data multiplexing in industrial safety risk management and control are solved, accurate management and control and intelligent interaction are realized, and support is provided for industrial safety guarantee.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial safety management, and particularly relates to a multi-source fusion industrial safety risk management and intelligent interaction method and system. BACKGROUND

[0002] Industrial safety management is a key field for ensuring production environment safety and reducing accident risks, and has important significance for enterprise operation and social stability. With the complication of industrial scenes, safety risk management needs to cover the whole chain from plant planning to personnel management, and the linkage between industrial control systems as the core hub of industrial production and the safety management system directly affects the real-time and effectiveness of risk management. Any oversight in any link may lead to serious consequences. However, the current safety management method lacks sufficient integration with industrial control systems, and has significant defects in dynamic and interactive aspects, which is difficult to meet the real-time, accurate and intelligent needs of modern industry, and there is an urgent need for technological innovation to fill this gap.

[0003] Existing safety management methods usually rely on static risk identification or paper-based safety training materials, which are difficult to adapt to the dynamic changes of risks in the plant. For example, traditional risk identification boards can only provide fixed information and cannot update risk levels or prompt content in real time according to the real-time state of the region. At the same time, the visitor management system lacks correlation with risk data, and when visitors enter a specific area, it is difficult to quickly obtain targeted safety guidelines. This static and fragmented approach results in a lag in the transmission of safety information, increasing the likelihood of personnel misoperation or ignoring risks.

[0004] Therefore, how to combine dynamic risk information with spatial positioning technology in a complex industrial scene and provide real-time and targeted safety guidance for different personnel through intelligent interaction has become a key problem in the field of safety management. SUMMARY

[0005] The present application provides a multi-source fusion industrial safety risk management and intelligent interaction method and system to solve the problems of inaccurate guidance, high path risk and difficult data reuse in industrial safety risk management and control, achieve accurate management and intelligent interaction, and provide support for industrial safety protection.

[0006] In a first aspect, to solve the above technical problems, the present application provides a multi-source fusion industrial safety risk management and intelligent interaction method, comprising: constructing a dynamic risk level distribution map of an industrial scene based on plant equipment operation parameters, environmental monitoring data, production operation data and historical safety data; determining the current accurate position of the target personnel and extracting real-time risk information of the region and the planned path; generating a personalized safety guidance sequence according to the target personnel and real-time risk information; superimpose the personalized safety guidance sequence to the personnel device display interface through augmented reality technology, generate a dynamic navigation path containing risk prompts and risk avoidance actions corresponding to each node of the path; If the risk intensity value of any node in the dynamic navigation path exceeds the preset risk threshold, a path optimization algorithm is used to recalculate a backup route and obtain a low-risk alternative navigation sequence; The updated risk information is extracted from the low-risk alternative navigation sequence and returned to the dynamic risk level distribution map to correct the regional risk data, synchronize the matching rules of the preset safety guidance generation model, and form refined safety management data; The personalized safety guidance sequence includes risk type corresponding protection prompts, navigation node risk avoidance instructions, and safety guidance.

[0007] In an optional embodiment, the dynamic risk level distribution map of the industrial scene is constructed based on plant equipment operation parameters, environmental monitoring data, production operation data, and historical safety data, including: Collect plant equipment operation parameters, environmental monitoring data, production operation data, and historical safety data, and perform cleaning, deduplication, and format standardization processing on various types of data to form a unified format risk analysis data set; According to the preset risk identification rules, extract the risk factors of each region of the plant from the risk analysis data set, and determine the risk type and influence range; Quantitative analysis of regional risk factors is performed through a risk assessment algorithm to determine the risk intensity value and risk level of the corresponding region; Map the risk type, risk intensity value, and risk level of each region to the plant space coordinates to generate a dynamic risk level distribution map of the industrial scene.

[0008] In an optional embodiment, the current accurate position of the target personnel is determined, and real-time risk information of the region and the planned path is extracted, including: Determine the current accurate position of the target personnel through noise reduction processing and coordinate calibration of the positioning terminal; Based on the work task or preset destination of the target personnel, define a planned path with the current accurate position as the starting point to determine the planned path range; Match the current accurate position and planned path range with the spatial coordinate system of the dynamic risk level distribution map to locate the corresponding physical region; According to the physical region, extract the matching risk type, risk intensity value, and real-time state data from the dynamic risk level distribution map to form the real-time risk information.

[0009] In an alternative embodiment, the generating of the personalized safety guidance sequence according to the target personnel and real-time risk information comprises: extracting the identity information and corresponding access rights and operation qualification data of the target personnel, and determining the adapted safety guidance reference range; analyzing the risk type, risk intensity value and influence area in the real-time risk information, matching the preset risk response rule library, and under the constraint of the safety guidance reference range, screening out the protection prompt and avoidance instruction corresponding to the current risk; adapting and checking the personnel capability level reflected by the safety guidance reference range with the screened protection prompt and avoidance instruction, adjusting the detail level of the protection prompt according to the personnel operation qualification level, and determining the execution priority of the avoidance instruction in combination with the risk intensity value and the personnel access rights; integrating the adapted protection prompt, path node risk avoidance instruction and identity exclusive safety instruction to form a complete personalized safety guidance sequence.

[0010] In an alternative embodiment, the superimposing of the personalized safety guidance sequence to the personnel device display interface through augmented reality technology to generate a dynamic navigation path containing risk prompts and risk avoidance actions corresponding to each node of the path comprises: calling the augmented reality technology to associate and map the risk information in the personalized safety guidance sequence with the spatial positioning data of the personnel device, establishing the coordinate correspondence relationship between the guidance content and the physical space; based on the coordinate correspondence relationship, superimposing the risk prompt identifier in the personnel device display interface, generating a dynamic navigation path with the current position as the starting point and the target position as the end point, and synchronously labeling the distribution nodes of the risk area; extracting the path node risk avoidance instruction in the personalized safety guidance sequence, and converting it into a visual node action instruction icon, and associating it to the corresponding path node; integrating the superimposed risk prompt identifier, dynamic navigation path and node action instruction icon to form a real-time interactive risk avoidance action.

[0011] In an alternative embodiment, the re-computing of the backup route by using a path optimization algorithm to obtain a low-risk alternative navigation sequence comprises: under the constraint of the current risk area distribution and dynamic risk level, calling the path optimization algorithm to perform global scanning on the original planned path, marking the path segments with risk intensity exceeding the standard, and outputting the marking result; based on the marking result, taking the avoidance of high-risk areas as the core target, combining the personnel device positioning data and the physical space topology structure, and generating potential backup routes; The potential alternate route is subjected to risk value quantitative evaluation, and a route with the lowest risk intensity and meeting the access permission is reserved as an optimal alternate route; The screened optimal alternate route is disassembled into continuous path nodes, and the risk prompts and access instructions corresponding to each node are associated to form a low-risk alternative navigation sequence.

[0012] In an optional implementation, the updated risk information is extracted from the low-risk alternative navigation sequence and fed back to the dynamic risk level distribution map to correct the regional risk data and synchronously adjust the matching rules of the preset safety guidance generation model to form refined safety management data, including: The real-time risk state changes of the path nodes in the low-risk alternative navigation sequence are analyzed to extract updated risk type, influence range and intensity fluctuation data as updated risk data output; The updated risk data is fed back to the dynamic risk level distribution map for comparison and verification to correct the risk level parameters and spatial distribution identifiers of the corresponding region; Based on the updated regional risk data, the matching weights of the risk type and response rules in the safety guidance generation model are adjusted to optimize the guidance output logic under different risk intensities; The corrected regional risk data and the adjusted matching weights of the response rules are integrated to form refined safety management data.

[0013] In an optional implementation, after the refined safety management data is formed, it further includes: The execution feedback data of the personnel on the personalized safety guidance sequence is collected; wherein the execution feedback data includes the understanding degree of the guidance content, the completion degree of the avoidance action and the risk avoidance effect; The feedback data and the refined safety management data are associated and analyzed to optimize the expression mode of the protection prompt and the trigger condition of the avoidance instruction in the safety guidance generation model.

[0014] In a second aspect, the present application also provides a multi-source fusion industrial safety risk management and intelligent interaction system, including: A risk map construction module: based on the plant equipment operation parameters, environmental monitoring data, production operation data and historical safety data, a dynamic risk level distribution map of the industrial scene is constructed; A risk information extraction module: determines the current accurate position of the target personnel and extracts the real-time risk information of the region and the planned path; A safety guidance generation module: generates a personalized safety guidance sequence according to the target personnel and real-time risk information; Real scene navigation generation module: superimpose the personalized safety guidance sequence to the personnel device display interface through augmented reality technology, generate a dynamic navigation path containing risk prompts and the risk avoidance actions corresponding to each node of the path; Low-risk path generation module: if the risk intensity value of any node in the dynamic navigation path exceeds the preset risk threshold, a path optimization algorithm is used to recalculate the backup route, and a low-risk alternative navigation sequence is obtained; Refined number integration formation module: extract the updated risk information from the low-risk alternative navigation sequence, and return it to the dynamic risk level distribution map to correct the regional risk data, synchronously adjust the matching rules of the preset safety guidance generation model, and form refined safety management data; The personalized safety guidance sequence includes risk type corresponding protection prompts, navigation node risk avoidance instructions and safety guidance.

[0015] In an optional embodiment, further comprising: Feedback detection module: collect personnel execution feedback data of the personalized safety guidance sequence; wherein the execution feedback data includes the understanding degree of the guidance content, the completion degree of the avoidance action and the risk avoidance effect; Optimization and improvement module: associate and analyze the feedback data and the refined safety management data to optimize the expression mode of the protection prompts and the trigger condition of the avoidance instructions in the safety guidance generation model.

[0016] Compared with the prior art, the present application has the following beneficial effects: (1) The dynamic risk level distribution map is constructed by fusing plant equipment operation parameters, environmental monitoring data and other multi-source data, which can comprehensively cover the risk influencing factors of industrial scenes, avoid the one-sidedness of single data source information, adapt to the characteristics of dynamic changes of plant risk with production state, improve the completeness and real-time performance of risk distribution presentation, and lay a data foundation for subsequent accurate extraction of risk information.

[0017] (2) The real-time risk information of the region and the planned path of the target personnel is extracted by combining the current position of the target personnel with the dynamic risk level distribution map, the accurate association of personnel position and risk data is realized, the risk information is avoided to be disconnected with the actual scene of the personnel, the spatial position difference of different personnel is adapted, and the personalized safety guidance generation is provided with a targeted basis.

[0018] (3) The personalized safety guidance sequence is generated based on the characteristics of the target personnel and the real-time risk information, the guidance content can be dynamically adjusted according to the personnel work qualification, the access permission and the risk scene, the problem of lack of adaptability of general guidance is avoided, the accuracy of safety guidance is improved, and the personnel are effectively guided to standardize risk avoidance.

[0019] (4) When the navigation path node risk exceeds the standard, the low-risk alternative navigation sequence is recalculated through the path optimization algorithm to dynamically avoid the sudden high-risk area, avoid the limitation that the fixed path cannot respond to the risk change, adapt to the dynamic adjustment demand of the risk threshold, and ensure the safety and timeliness of personnel access and risk avoidance.

[0020] (5) The updated risk information in the alternative navigation sequence is extracted, the dynamic risk level distribution map is corrected, and the guidance model rule is adjusted to form refined safety management data, realize the iterative optimization of risk data and model rules, avoid the lag of data and model from the actual risk change, build a risk control closed loop, and improve the practical value of data and model.

[0021] (6) The execution feedback data of personnel on the guidance sequence is collected and analyzed in association with the refined safety management data to optimize the guidance model. The protection prompt expression and avoidance instruction trigger condition can be adjusted according to the actual understanding and execution of personnel to avoid the lack of application support in model optimization, improve the easy understanding of guidance content and the accuracy of instruction triggering, and enhance the practicality of guidance. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is a flowchart of a multi-source fusion industrial safety risk control and intelligent interaction method provided by an embodiment of the present application; Figure 2 is a structural schematic diagram of a multi-source fusion industrial safety risk control and intelligent interaction system provided by an embodiment of the present application. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0024] Referring to Figure 1 , the embodiment of the present application provides a multi-source fusion industrial safety risk control and intelligent interaction method, comprising the following steps: S11, based on the plant equipment operation parameters, environmental monitoring data, production operation data and historical safety data, a dynamic risk level distribution map of the industrial scene is constructed; S12, the current accurate position of the target personnel is determined, and the real-time risk information of the area and the planned path is extracted; S13, according to the target personnel and real-time risk information, a personalized safety guidance sequence is generated; S14, superimpose the personalized safety guidance sequence to the personnel device display interface through augmented reality technology to generate a dynamic navigation path containing risk prompts and corresponding risk avoidance actions for each node of the path; S15, if the risk intensity value of any node in the dynamic navigation path exceeds the preset risk threshold value, a backup route is recalculated using a path optimization algorithm to obtain a low-risk alternative navigation sequence; S16, extract updated risk information from the low-risk alternative navigation sequence and return it to the dynamic risk level distribution map to correct the regional risk data, synchronously adjust the matching rules of the preset safety guidance generation model, and form refined safety management data; In step S11, based on plant equipment operation parameters, environmental monitoring data, production operation data and historical safety data, a dynamic risk level distribution map of the industrial scene is constructed.

[0025] In an embodiment, when constructing the dynamic risk level distribution map of the industrial scene, first, multi-source data acquisition is completed through the distributed data acquisition terminal and the industrial control system (such as the SCADA system) deployed in the plant: equipment operation parameters are collected by PLC controllers and synchronized to the industrial control system, covering 12 types of parameters such as reaction kettle temperature (collection accuracy ±0.1℃), compressor pressure (collection accuracy ±0.01MPa); environmental monitoring data is obtained by a sensor array and uploaded to the industrial control system in real time, including combustible gas concentration (0-100% LEL), dust concentration (0-100mg / m 3 ), etc.; production operation data is synchronized to the industrial control system through the MES system, and historical safety data is retrieved from the enterprise safety management platform associated with the industrial control system. After collection, 3σ criterion is used to eliminate outliers, de-duplication is performed through hash algorithm, standardized fields and units are referred to GB / T22239-2019, and a risk analysis data set of "timestamp-region number-parameter type-value" is formed.

[0026] According to the preset risk identification rules (such as "reaction kettle temperature > 140℃ and continuous 5min is determined as high temperature risk"), risk factors of each region are extracted, and the risk type and influence range (defined by 5m×5m grid unit) are determined. After determining the weight of each risk factor by AHP, the regional comprehensive risk intensity value is quantitatively calculated by the following formula: , wherein R is the regional comprehensive risk intensity value (0-100), n is the number of risk factors, w i is the weight of the ith risk factor, V i is the real-time monitoring value, V min,i , V max,iRespectively, the lower limit of the safety threshold and the upper limit of the dangerous threshold. According to the R value, four risk levels are divided (level I: 0-20, level II: 21-40, level III: 41-70, level IV: 71-100), and finally associated with the plant space coordinates (with the northwest corner as the origin) mapping, through ArcGIS to generate a dynamic risk level distribution map marked with red, orange, yellow and green, updated at a frequency of 1 min / second.

[0027] In step S12, the current accurate position of the target person is determined, and the real-time risk information of the area and the planned path is extracted.

[0028] In one embodiment, based on the dynamic risk level distribution map constructed above and synchronized to the industrial control system, when obtaining the current position information of the target personnel and extracting real-time risk information, first, the target personnel is equipped with a UWB positioning terminal (positioning accuracy ± 0.3 m), which interacts with 16 positioning base stations deployed in the factory area, collects position signals 3 times per second and uploads them to the industrial control system; for the base station interference noise existing in the collected position signals, a Kalman filter algorithm is used for noise reduction processing to remove the random fluctuation components in the signals, and then combined with the preset coordinate calibration points in the factory area (1 point every 50 m, a total of 20 points), the coordinates of the signals after noise reduction are calibrated by the triangulation method, and finally the accurate current position of the target personnel (such as coordinates (32.5 m, 46.8 m)) is determined and synchronized to the industrial control system. If the target personnel is an operator, based on the work task assigned in his MES system (such as "3# reaction kettle inspection"), the destination is determined to be the area where the 3# reaction kettle is located (coordinate range (50m-60m, 70m-80m)); if it is a visitor, based on the preset destination registered (such as "office building 3rd floor conference room"), the accurate current position is taken as the starting point, and the Dijkstra algorithm is used to determine the planned path, with a path width of 2 m, forming a planned path range extending 1 m to both sides of the path centerline. The determined accurate current position coordinates and the coordinate boundaries of the planned path range are accurately matched with the spatial coordinate system used by the dynamic risk level distribution map in the industrial control system, which is "the northwest corner of the factory area as the origin, x axis east-west, y axis south-north", to locate the physical area where the target personnel is currently located (such as Area A of No. 2 production workshop) and the physical area covered by the planned path (such as the corridor of No. 2 workshop and the east corridor of raw material storage area). Then, the risk information corresponding to the above matched physical areas is extracted from the dynamic risk level distribution map of the industrial control system, including the "mechanical failure risk" of Area A of No. 2 production workshop (risk intensity value 35, medium risk), the "combustible gas leakage risk" of the east corridor of raw material storage area (risk intensity value 28, medium risk), and the real-time state data of each area risk (such as "mechanical failure risk duration 12 min" "combustible gas concentration real-time value 18% LEL"), which are integrated in the format of "area number-risk type-risk intensity value-real-time state" to form the real-time risk information of the target personnel and fed back to the industrial control system.

[0029] In step S13, a personalized safety guidance sequence is generated according to the target personnel and real-time risk information.

[0030] In one embodiment, based on the above-mentioned target personnel accurate position and real-time risk information obtained and fed back to the industrial control system, when generating a personalized safety guidance sequence, first, the identity information of the target personnel (such as “work personnel Zhang, work number 0012”) is extracted through the plant personnel management system associated with the industrial control system, the corresponding access permission data (allowed to enter the production areas such as No. 2 production workshop, raw material storage area, prohibited to enter the core area of the hazardous chemical warehouse) and work qualification data (holds a secondary equipment inspection qualification, has a basic judgment ability of mechanical failure) are synchronously called, and the adaptive safety guidance reference range is determined accordingly— for the equipment inspection scene, the basic protection and avoidance guidance of mechanical failure and gas leakage risk are included, and the high-difficulty guidance content such as hazardous chemical professional disposal which exceeds the qualification is excluded. Then, the real-time risk information in the industrial control system is analyzed to determine the “mechanical failure risk” (risk intensity value 35, medium risk, impact area No. 2 production workshop A area) and “combustible gas leakage risk” (risk intensity value 28, medium risk, impact area east side passage of raw material storage area), and the preset risk response rule library (the rule library includes 120 risk response rules, associated with risk type, intensity and corresponding disposal scheme) in the industrial control system is matched, and under the constraint of the above-mentioned safety guidance reference range, the corresponding protection prompt (such as “wear anti-punch gloves, safety helmet, and keep a distance of more than 1.5 m from the fault equipment” for mechanical failure risk, and “wear portable combustible gas detector, prohibit the use of open flame” for combustible gas leakage risk) and avoidance instruction (such as “avoid the north side of No. 2 workshop A area fault equipment area, and detour the south side passage”) are screened out. Then, the personnel capability level reflected by the safety guidance reference range (secondary inspection qualification, with basic judgment ability) is adapted and verified with the screened protection prompt and avoidance instruction: according to the work qualification level, the detailed level of mechanical failure protection prompt is adjusted to “basic operation level” (omitting professional fault troubleshooting steps, only retaining protection and avoidance points), and combined with the risk intensity value (35>28) and the personnel access permission (both are allowed to enter the area), it is determined that the avoidance instruction execution priority of “avoiding the fault equipment area” is higher than that of “leakage area access protection”. Finally, the adapted protection prompt, path node risk avoidance instruction, and identity exclusive safety instructions (such as “you have a secondary inspection qualification, can observe the appearance of the fault equipment, and are prohibited from disassembling the machine”) for the personnel qualification are integrated to form a complete personalized safety guidance sequence, which is presented in the logic of “risk type-protection prompt-avoidance instruction-exclusive instruction” and stored in the industrial control system.

[0031] In step S14, the personalized safety guidance sequence is superimposed to the personnel equipment display interface through augmented reality technology to generate a dynamic navigation path containing risk prompts and risk avoidance actions corresponding to each node of the path.

[0032] In one implementation, based on the personalized safety guidance sequence generated and stored in the industrial control system, when interactive presentation is achieved through augmented reality (AR) technology, the AR development engine built into the AR smart glasses (1920×1080 resolution, 52° field of view) worn by the target personnel is first invoked. Information such as "mechanical failure risk" and "combustible gas leakage risk" in the personalized safety guidance sequence is retrieved from the industrial control system and associated with the personnel spatial positioning data (such as coordinates (32.5m, 46.8m) and movement direction angle of 30°) uploaded in real time by the UWB positioning terminal. Based on the unified spatial coordinate system of the factory area, a precise coordinate correspondence between each guidance content and the physical space is established (such as binding the protection prompt of "faulty equipment on the north side of area A of workshop No. 2" to the area range of coordinates (30m, 45m) - (35m, 50m). Based on this coordinate correspondence, risk warning signs are overlaid on the display interface of the AR smart glasses: orange triangular warning icons (2cm on each side) are marked for medium-risk mechanical failure areas, with the text "Mechanical failure, keep a distance of 1.5m" displayed; yellow circular warning icons (2cm in diameter) are marked for flammable gas leak areas, with the text "Flammable gas leak, no open flames allowed" displayed simultaneously. Meanwhile, starting from the personnel's current location (32.5m, 46.8m) and ending at the area where reactor No. 3 is located (55m, 75m), a dynamic navigation path is generated by the AR engine. The path is presented as a solid green line (0.5cm wide), and the path direction is updated every 1 second based on the personnel's real-time location. Risk area distribution nodes are simultaneously marked on the path (e.g., a leak risk sign is marked at the node of the passageway on the east side of the raw material storage area). Subsequently, the risk avoidance instructions for path nodes in the personalized safety guidance sequence were extracted. "Detour through the south passage of Area A in Workshop 2" was transformed into a right-turning arrow icon (1.5cm x 1cm), and "Wear a combustible gas detector" was transformed into a cartoon icon of the instrument (2cm x 2cm). These visual icons were then associated with corresponding path nodes (e.g., the turning arrow was bound to the path inflection point at coordinates (32.5m, 50m), and the instrument icon was bound to the raw material storage area entrance node (40m, 60m)). Finally, the superimposed risk warning signs, dynamic navigation paths, and node action guidance icons were integrated to form a real-time interactive risk avoidance action: when personnel are within 1m of the risk area boundary (near the faulty equipment area), the AR interface automatically enlarges the orange warning icon and emits a slight vibration reminder; when reaching a turning node, the arrow icon flashes to indicate the direction of travel; before entering the leak risk area, the instrument wearing icon pops up with a voice prompt "Please wear a combustible gas detector," while simultaneously feeding back the personnel's interaction status to the industrial control system in real time, ensuring that personnel can intuitively obtain and execute risk avoidance operations.

[0033] In step S15, if the risk intensity value of any node in the dynamic navigation path exceeds the preset risk threshold, the path optimization algorithm is used to recalculate the backup route, and a low-risk alternative navigation sequence is obtained.

[0034] In an embodiment, based on the above-mentioned AR dynamic navigation path generation result, if the industrial control system real-time monitors that the combustible gas leakage risk intensity value of the east passage node of the raw material storage area in the original planned path suddenly rises to 42 (exceeding the preset medium risk threshold 40), the path optimization process is started. Taking the risk area distribution (2nd workshop A area mechanical failure medium risk, east passage of raw material storage area high risk) of the dynamic risk level distribution map in the industrial control system as the constraint condition, the improved A* path optimization algorithm is called to perform global scanning on the original planned path (32.5m, 46.8m)→(40m, 60m)→(55m, 75m), and by traversing the risk intensity values of each 5m×5m grid cell on the path, the east passage (40m, 60m)→(45m, 70m) segment of the raw material storage area is marked as the path segment with excessive risk intensity. Based on the marking result, taking avoiding high-risk areas as the core target, combining the real-time position data (32.5m, 46.8m) of personnel obtained by the UWB positioning terminal and uploaded to the industrial control system and the physical space topology structure (such as passage width, obstacle distribution, regional connectivity) of the factory area, 3 potential backup routes are generated: route 1 is (32.5m, 46.8m)→(32.5m, 70m)→(55m, 75m), route 2 is (32.5m, 46.8m)→(25m, 60m)→(55m, 75m), and route 3 is (32.5m, 46.8m)→(38m, 50m)→(50m, 50m)→(55m, 75m). Then the risk value of the 3 routes is quantitatively evaluated, the weighted summation method is used to calculate the comprehensive risk value (risk intensity value×path length proportion, the path length proportion is calculated based on the original path length) of each route, it is obtained that the comprehensive risk value of route 1 is 22, the comprehensive risk value of route 2 is 25, and the comprehensive risk value of route 3 is 18, at the same time, the route reachability (no closed passage, satisfying the 2m passing width) and the path length rationality (the length of route 3 increases by 15% compared with the original path, which is within the acceptable range) are verified, and the route 3 with the lowest risk intensity and satisfying the personnel passing permission (all allowing to enter the area) is preferentially retained. Finally, the route 3 is disassembled into 3 continuous path nodes (32.5m, 46.8m)→(38m, 50m), (38m, 50m)→(50m, 50m), (50m, 50m)→(55m, 75m), respectively associated with the risk prompts (such as the (38m, 50m) node is associated with “pay attention to the right side equipment running noise, wear earplugs”) and the passing instructions (such as the (50m, 50m) node is associated with “straight into the reaction kettle area, speed limit 5km / h”) of each node, and integrated into a low-risk alternative navigation sequence and uploaded to the industrial control system.

[0035] In step S16, the updated risk information is extracted from the low-risk alternative navigation sequence and returned to the dynamic risk level distribution map to correct the regional risk data and synchronously adjust the matching rules of the preset safety guidance generation model to form refined safety management data.

[0036] In an embodiment, based on the low-risk alternative navigation sequence (route 3), the real-time risk state changes of each path node are analyzed: the node (38m, 50m) adds a "device operation noise risk" (influence range 3m x 3m, intensity value 22), the intensity of the node (50m, 50m) "mechanical vibration risk" increases from 18 to 25, and the intensity of the high-risk area of the east passage of the raw material storage area fluctuates from 42 to 45 to 43 and the range shrinks to 4m x 6m. These updated data are returned to the dynamic risk level distribution map of the industrial control system, and after comparison and verification, the corresponding regional risk level parameters (such as (50m, 50m) still being low risk) and spatial distribution marks (noise risk marked with light blue circles, and the red high-risk range is reduced) are corrected to ensure that the dynamic risk level distribution map is updated synchronously.

[0037] Based on the corrected regional risk data, the matching weight of the "risk type-response rule" of the safety guidance generation model in the industrial control system is adjusted, and the calculation formula is as follows: wherein, is the matching weight of the jth risk and the kth rule after adjustment, w j,k is the initial weight, a is the adjustment coefficient (0.2), F j is the execution feedback score (0-100) of the jth risk guidance, F avg is the average score, F max =100. For example, the matching weight of "device operation noise risk" and "wearing earplugs" is adjusted from 0.5 to 0.8, and the guidance output logic in the risk intensity range of 20-30 is optimized (an "easy protection prompt" is added). Finally, the corrected risk data and the adjusted weight are integrated to form refined safety management data in the standardized format of GB / T22239-2019, which is stored in the distributed database of the plant safety management platform to provide data support for subsequent processes.

[0038] It is worth noting that after step S16, the following steps are also included: S17, collecting execution feedback data of personnel on the personalized safety guidance sequence; wherein the execution feedback data includes the understanding degree of the guidance content, the completion degree of the avoidance action, and the risk avoidance effect; S18, correlating and analyzing the feedback data with the refined safety management data to optimize the expression mode of the protection prompt and the trigger condition of the avoidance instruction in the safety guidance generation model.

[0039] In one embodiment, after the refined safety management data is formed, a safety guidance optimization closed-loop process is started: first, the execution feedback data is collected through the built-in interaction module of the AR smart glasses worn by the target personnel, wherein the understanding degree of the guidance content is comprehensively judged by the real-time voice feedback of the personnel (such as “completely understand” “partially blurred”) and the interface operation response speed (the click confirmation time is less than 3s, which is considered to understand smoothly), and is divided into three levels of “completely understand” “partially understand” “not understand”; the completion degree of avoidance action is monitored in real time through AR visual recognition technology, such as whether the personnel wear anti-throw gloves according to the guidance (the recognition accuracy is greater than or equal to 95%), whether they walk around to the designated channel (the position deviation is less than 1m, which is considered to be completed), so as to quantify the completion rate (the actual number of completed actions / the number of actions that should be completed); the risk avoidance effect is evaluated by comparing the actual risk triggering situation (such as whether to enter a high-risk area, whether to trigger an equipment abnormal warning) of the personnel passing through the area with the expected avoidance target, and is evaluated as “effective avoidance” “partial avoidance” “non-avoidance”; all feedback data is uploaded to the industrial control system in real time.

[0040] The collected feedback data (such as Zhang’s understanding of the mechanical failure protection prompt is “completely understand”, the avoidance action completion rate is 85%, and the risk avoidance effect is “effective avoidance”) is associated with the corresponding area risk parameters (the mechanical failure risk intensity of No. 2 workshop A area is 35) of the refined safety management data (extracted from the distributed database through the / api / safety / refined-data interface) in the industrial control system and the guidance model matching rule (the matching weight of mechanical failure and anti-throw protection is 0.7) for correlation analysis: it is found that the prompt of “wearing a portable combustible gas detector” is marked as “partially understand” in 30% of the feedback, because the word “portable” is easy to be confused with fixed detection equipment; at the same time, the avoidance instruction of “walking around the south side channel” triggers a delay of about 2s when the risk intensity is greater than or equal to 30. According to this, the safety guidance generation model in the industrial control system is optimized: “portable combustible gas detector” is adjusted to “handheld combustible gas detector worn on the body” to improve the clarity of the statement; the avoidance instruction trigger threshold when the mechanical failure risk intensity is greater than or equal to 25 is advanced, and the response delay is shortened to within 0.5s, to ensure that the subsequent generated safety guidance is more in line with the understanding habit of personnel and the risk disposal time efficiency requirement.

[0041] In summary, the present application realizes the whole-chain management and intelligent interaction of industrial safety risks by constructing the complete technical logic of "multi-source data fusion perception-dynamic risk precise mapping-personalized guidance intelligent generation-real-time interaction and path optimization-data closed-loop iteration". Specifically, first, the dynamic risk level distribution map is constructed by fusing multi-source data such as plant equipment operation and environmental monitoring, solving the problem that traditional static management cannot reflect the spatio-temporal changes of risks; second, real-time risk information is extracted based on personnel positioning, and personalized guidance is generated based on personnel qualifications, and the intuitive fusion of risk prompts and physical scenes is realized through augmented reality technology, solving the problems of risk information transmission lag and disconnection with the actual scene; when the path risk exceeds the standard, a low-risk alternative route is generated through algorithm, and the updated risk information is returned to correct the distribution map and optimize the guidance model, and the closed-loop mechanism of "perception-decision-execution-feedback" is formed through continuous iteration combined with personnel execution feedback.

[0042] This technical logic effectively solves the core problems of risk dynamic presentation deficiency, low guidance accuracy and poor interaction responsiveness in existing methods, realizes the real-time presentation of risk information, the accurate push of personalized safety guidance and the intelligent optimization of dynamic path, and significantly improves the timeliness, pertinence and effectiveness of risk management in industrial scenes, providing strong support for personnel safety protection and efficient production and operation in complex industrial environments.

[0043] Reference Figure 2 The second embodiment of the application provides a multi-source fusion industrial safety risk management and intelligent interaction system, comprising: A risk map construction module: based on plant equipment operation parameters, environmental monitoring data, production operation data and historical safety data, a dynamic risk level distribution map of an industrial scene is constructed; A risk information extraction module: determines the current accurate position of a target person and extracts real-time risk information of the area and planned path; A safety guidance generation module: generates a personalized safety guidance sequence according to the target person and real-time risk information; A real scene navigation generation module: superimposes the personalized safety guidance sequence to the personnel equipment display interface through augmented reality technology to generate a dynamic navigation path containing risk prompts and risk avoidance actions corresponding to each node of the path; A low-risk path generation module: if the risk intensity value of any node in the dynamic navigation path exceeds the preset risk threshold, a path optimization algorithm is used to recalculate the backup route to obtain a low-risk alternative navigation sequence; A refined number integration forming module: extracts the updated risk information from the low-risk alternative navigation sequence, returns it to the dynamic risk level distribution map to correct the regional risk data, synchronously adjusts the matching rules of the preset safety guidance generation model, and forms refined safety management data; The personalized safety guidance sequence includes a protection prompt corresponding to a risk type, a navigation node risk avoidance instruction, and safety guidance.

[0044] The multi-source fusion industrial safety risk management and intelligent interaction system further includes: A feedback detection module collects execution feedback data of the personnel on the personalized safety guidance sequence, wherein the execution feedback data includes understanding of the guidance content, completion of the avoidance action, and risk avoidance effect. An optimization and improvement module analyzes the feedback data in association with refined safety management data, optimizes a representation manner of the protection prompt and a trigger condition of the avoidance instruction in the safety guidance generation model.

[0045] It should be noted that the multi-source fusion industrial safety risk management and intelligent interaction system provided by the embodiments of the present application is used to execute all process steps of the multi-source fusion industrial safety risk management and intelligent interaction method of the above embodiments, and the working principles and beneficial effects of the two are one-to-one correspondence, so they will not be described again.

[0046] The above-described specific embodiments further detail the purpose, technical solutions, and beneficial effects of the present application. It should be understood that the above-described only specific embodiments of the present application and are not intended to limit the scope of protection of the present application. It is particularly pointed out that any modifications, equivalent replacements, improvements, etc. made by those skilled in the art within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A multi-source integrated industrial safety risk management and intelligent interaction method, characterized in that, include: Based on the plant's equipment operating parameters, environmental monitoring data, production operation data, and historical safety data, a dynamic risk level distribution map for industrial scenarios is constructed. Determine the precise current location of the target personnel and extract real-time risk information about the area and planned route; Based on the target personnel and real-time risk information, a personalized safety guidance sequence is generated; The personalized safety guidance sequence is overlaid onto the personnel and equipment display interface using augmented reality technology to generate a dynamic navigation path with risk warnings and risk avoidance actions corresponding to each node of the path. If the risk intensity value of any node in the dynamic navigation path exceeds the preset risk threshold, the path optimization algorithm is used to recalculate the backup route and obtain a low-risk alternative navigation sequence. Updated risk information is extracted from the low-risk alternative navigation sequence and sent back to the dynamic risk level distribution map to correct regional risk data. The matching rules of the preset safety guidance generation model are adjusted synchronously to form refined safety management data. The personalized safety guidance sequence includes protection prompts corresponding to risk types, navigation node risk avoidance instructions, and safety guidance.

2. The multi-source fusion industrial safety risk management and intelligent interaction method according to claim 1, characterized in that, The dynamic risk level distribution map of the industrial scenario is constructed based on the plant's equipment operating parameters, environmental monitoring data, production operation data, and historical safety data, including: Collect equipment operating parameters, environmental monitoring data, production operation data and historical safety data in the plant area, clean and deduplicate the various types of data and standardize the format to form a risk analysis dataset with a unified format; Based on the preset risk identification rules, risk factors for each area of ​​the factory are extracted from the risk analysis dataset to clarify the risk type and its scope of impact. The risk factors in each region are quantitatively analyzed using risk assessment algorithms to determine the risk intensity value and risk level of the corresponding region. By mapping the risk type, risk intensity value, and risk level of each region to the spatial coordinates of the factory area, a dynamic risk level distribution map of the industrial scenario is generated.

3. The multi-source fusion industrial safety risk management and intelligent interaction method according to claim 1, characterized in that, The process of determining the precise current location of the target personnel and extracting real-time risk information about the area and planned route includes: The location signal is collected by the positioning terminal, and after noise reduction and coordinate calibration, the current precise location of the target person is determined. Based on the target personnel's work tasks or preset destinations, a planned path is delineated starting from the current precise location, and the range of the planned path is determined. Match the current precise location and planned path range with the spatial coordinate system of the dynamic risk level distribution map to locate the corresponding physical area; Based on the physical region, the matching risk type, risk intensity value, and real-time status data are extracted from the dynamic risk level distribution map and integrated to form the real-time risk information.

4. The multi-source fusion industrial safety risk management and intelligent interaction method according to claim 1, characterized in that, The step of generating a personalized safety guidance sequence based on the target personnel and real-time risk information includes: Extract the identity information of the target personnel and their corresponding access permissions and work qualification data to determine the appropriate safety guidance benchmark range; The risk type, risk intensity value and affected area in the real-time risk information are analyzed, matched with the preset risk response rule base, and under the constraints of the safety guidance benchmark range, the protection prompts and avoidance instructions corresponding to the current risk are filtered out. The personnel capability level reflected by the safety guidance benchmark range is matched and verified with the selected protection prompts and avoidance instructions. The level of detail of the protection prompts is adjusted according to the personnel's work qualification level, and the execution priority of the avoidance instructions is determined by combining the risk intensity value and personnel access rights. The integrated protection prompts, path node risk avoidance instructions, and identity-specific security instructions form a complete personalized security guidance sequence.

5. The multi-source fusion industrial safety risk management and intelligent interaction method according to claim 1, characterized in that, The process of overlaying the personalized safety guidance sequence onto the personnel's device display interface using augmented reality technology to generate a dynamic navigation path with risk warnings and corresponding risk avoidance actions for each node on the path includes: Augmented reality technology is used to associate and map the risk information in the personalized safety guidance sequence with the spatial positioning data of personnel and equipment, and to establish a coordinate correspondence between the guidance content and the physical space. Based on the coordinate correspondence, risk warning signs are overlaid on the personnel and equipment display interface, and a dynamic navigation path is generated with the current location as the starting point and the target location as the ending point, while the distribution nodes of the risk area are marked simultaneously. Extract the risk avoidance instructions from the path nodes in the personalized safety guidance sequence, convert them into visual node action guidance icons, and associate them with the corresponding path nodes; The integrated and superimposed risk warning signs, dynamic navigation paths, and node action guidance icons form a real-time interactive risk avoidance action.

6. The multi-source fusion industrial safety risk management and intelligent interaction method according to claim 1, characterized in that, The step of recalculating alternative routes using a path optimization algorithm to obtain low-risk alternative navigation sequences includes: Using the current risk area distribution and dynamic risk level as constraints, the path optimization algorithm is invoked to perform a global scan of the original planned path, marking the path segments with excessive risk intensity, and outputting the marking results. Based on the marking results, with the core objective of avoiding high-risk areas, potential alternative routes are generated by combining personnel and equipment positioning data with physical space topology. The potential backup routes are quantitatively assessed for risk, and the routes with the lowest risk intensity that meet the access rights are retained to obtain the optimal backup routes. The selected optimal alternative routes are broken down into continuous path nodes, and risk warnings and passage instructions corresponding to each node are associated with them, and integrated into a low-risk alternative navigation sequence.

7. The multi-source fusion industrial safety risk management and intelligent interaction method according to claim 1, characterized in that, The process of extracting updated risk information from the low-risk alternative navigation sequence, transmitting it back to the dynamic risk level distribution map to correct regional risk data, and simultaneously adjusting the matching rules of the preset safety guidance generation model to form refined safety management data includes: The real-time risk status changes of path nodes are analyzed from the low-risk alternative navigation sequence, and the updated risk type, impact range and intensity fluctuation data are extracted as the updated risk data output. The updated risk data is sent back to the dynamic risk level distribution map for comparison and verification, and the risk level parameters and spatial distribution labels of the corresponding areas are corrected. Based on the updated regional risk data, the matching weights of risk types and response rules in the safety guidance generation model are adjusted, and the guidance output logic under different risk intensities is optimized. By integrating the revised regional risk data and the matching weights of the adjusted response rules, refined safety management data is formed.

8. The multi-source fusion industrial safety risk management and intelligent interaction method according to claim 1, characterized in that, After refining the safety management data, the following is also included: The data collected includes feedback on the execution of personalized safety guidance sequences by the personnel; wherein the execution feedback data includes the degree of understanding of the guidance content, the degree of completion of avoidance actions, and the risk avoidance effect. By correlating and analyzing feedback data with refined safety management data, we can optimize the way protective prompts are expressed and the triggering conditions for avoidance instructions in the safety guidance generation model.

9. A multi-source integrated industrial safety risk management and intelligent interaction system, characterized in that, include: Risk Mapping Module: Based on plant equipment operating parameters, environmental monitoring data, production operation data, and historical safety data, a dynamic risk level distribution map of the industrial scenario is constructed. Risk information extraction module: Determines the precise current location of the target personnel and extracts real-time risk information about the area and planned route; Safety guidance generation module: Generates a personalized safety guidance sequence based on the target personnel and real-time risk information; Real-scene navigation generation module: Using augmented reality technology, the personalized safety guidance sequence is overlaid onto the personnel and equipment display interface to generate a dynamic navigation path with risk warnings and risk avoidance actions corresponding to each node of the path; Low-risk path generation module: If the risk intensity value of any node in the dynamic navigation path exceeds the preset risk threshold, the alternative route is recalculated using a path optimization algorithm to obtain a low-risk alternative navigation sequence. The refined data integration module extracts updated risk information from the low-risk alternative navigation sequence, transmits it back to the dynamic risk level distribution map to correct regional risk data, and simultaneously adjusts the matching rules of the preset safety guidance generation model to form refined safety management data. The personalized safety guidance sequence includes protection prompts corresponding to risk types, navigation node risk avoidance instructions, and safety guidance.

10. The multi-source fusion industrial safety risk management and intelligent interaction system according to claim 9, characterized in that, Also includes: Feedback Detection Module: Collects execution feedback data of personnel on personalized safety guidance sequences; wherein, the execution feedback data includes the degree of understanding of the guidance content, the degree of completion of avoidance actions, and the risk avoidance effect; Optimization and Improvement Module: Correlate and analyze feedback data with refined safety management data to optimize the expression of protection prompts and the triggering conditions of avoidance instructions in the safety guidance generation model.

Images