An emergency command and dispatch method and platform for industrial parks based on a large model

Abstract

This invention provides a method and platform for emergency command and dispatch in industrial parks based on a large model, comprising: identifying safety hazards in industrial parks based on visual technology; constructing hazard handling plans for the industrial parks using a large model; dividing the industrial parks into several emergency impact zones using the large model when an emergency occurs, constructing corresponding hazard handling plans for each emergency impact zone, collecting personnel retention information within each emergency impact zone, constructing evacuation and disposal plans for on-site personnel based on the hazard attributes of the corresponding emergency impact zone, identifying the deficiencies in the current emergency handling based on the on-site emergency data of the industrial parks, and generating corresponding emergency command commands to conduct on-site command. This significantly improves emergency response speed, decision-making accuracy, rescue success rate, and departmental coordination rate.

Description

Technical Field

[0001] This invention relates to the field of emergency response technology in industrial parks, and in particular to an emergency command and dispatch method and platform for industrial parks based on a large model. Background Technology

[0002] With the rapid development of industry, the number and scale of various industrial parks are constantly expanding. The chemical industry, encompassing multiple branches such as oil refining, metallurgy, pharmaceuticals, and coal chemicals, plays a crucial role in the overall industrial development. For different industrial parks, safety remains a key concern. To keep pace with the times, industrial parks need to continuously expand their scale and introduce new equipment, but this also brings more safety challenges. Therefore, many parks have implemented intensive management, deploying more personnel for safety supervision. While this achieves the goal of supervision, the results are unsatisfactory. Managers struggle to accumulate experience in daily management, making it difficult to respond effectively to emergencies when accidents occur. Therefore, there is an urgent need for a new method for managing the safety of industrial parks.

[0003] Therefore, this invention provides an emergency command and dispatch method and platform for industrial parks based on a large model. Summary of the Invention

[0004] This invention provides an emergency command and dispatch method and platform for industrial parks based on a large model, which significantly improves emergency response speed, decision-making accuracy, rescue success rate, and inter-departmental coordination rate.

[0005] This invention provides an emergency command and dispatch method for industrial parks based on a large model, comprising:

[0006] Step 1: Identify safety hazards in the industrial park based on visual technology, and construct a hazard mitigation plan for the industrial park using a large model;

[0007] Step 2: When an emergency occurs in the industrial park, the large model is used to divide the industrial park into several emergency impact areas, and a corresponding hazard disposal plan is constructed for each emergency impact area.

[0008] Step 3: Collect information on the number of people stranded within each of the aforementioned emergency impact areas, and construct an evacuation and disposal plan for the on-site personnel based on the hazard attributes of the corresponding emergency impact areas;

[0009] Step 4: Based on the on-site emergency data of the industrial park, determine the deficiencies in the current emergency response and generate corresponding emergency command commands to conduct on-site command.

[0010] In one feasible approach

[0011] Step 1 includes:

[0012] Step 11: Use visual recognition technology to extract features from the historical surveillance video of the industrial park to obtain several historical local features of the industrial park. Simultaneously compare the historical local features corresponding to the same area of ​​the park to obtain the local feature period corresponding to each area of ​​the park and construct the coarse feature change trend of each area of ​​the park.

[0013] Step 12: Use visual recognition technology to extract features from the real-time monitoring video of the industrial park to obtain several real-time local features of the industrial park. Use the coarse feature transformation trend to compare the corresponding real-time local features and extract the target real-time local features that are inconsistent with the coarse feature transformation trend.

[0014] Step 13: Locate the target park area corresponding to the real-time local features of the target in the real-time monitoring video, deduce the project parameters of the target park area based on the regional engineering projects of the target park area and the real-time local features of the target, and determine the safety hazards of the target park area;

[0015] Step 14: Input the safety hazards and the regional engineering projects into the large model to trace the source of the hazards, obtain the causes and locations of the hazards in the target park area, screen the corresponding initial hazard handling plans, decompose the regional engineering projects, determine the engineering parameters corresponding to the location of the hazards, and use the engineering parameters to adjust the data of the initial hazard handling plans to obtain the hazard handling plan for the industrial park.

[0016] In one feasible approach

[0017] Also includes:

[0018] Using the real-time local features, the regional engineering progress corresponding to the park area is determined, and the safety hazard database of the industrial park is updated according to the regional engineering progress.

[0019] In one feasible approach

[0020] Step 2 includes:

[0021] Step 21: Monitor the real-time video of the industrial park in real time. When an emergency occurs in the industrial park, perform frame-by-frame processing on the real-time video and determine the time point corresponding to each video frame. Convert each video frame into numerical data and combine the corresponding time points to obtain a multi-dimensional array corresponding to each video frame.

[0022] Step 22: Input each of the multidimensional arrays into the large model for data analysis based on the time sequence. Divide the industrial park into several emergency impact ranges according to the emergency danger locations and emergency danger attributes in the industrial park, and determine the range radius and danger level corresponding to each emergency impact range.

[0023] Step 23: Based on the emergency danger attributes, deduce several affected features corresponding to each emergency impact range, use a preset collaborative neural network to perform multi-angle learning and training on each affected feature, obtain several scene parameters corresponding to each affected feature, analyze the parameter logic between different scene parameters, determine the feature relationship between different affected features, and construct the corresponding feature tree.

[0024] Step 24: Construct the associated influence factor corresponding to each of the affected features based on the feature tree; use the large model to construct a corresponding first feature treatment scheme for the target affected feature with the largest associated influence factor; determine the causal treatment result of the first feature treatment scheme for each of the affected features based on the feature tree; and construct a second treatment scheme for the corresponding affected feature based on the causal treatment result.

[0025] Step 25: Obtain the handling plan corresponding to each of the affected features and construct the hazard handling plan for the industrial park. Find the hazard handling plan corresponding to each emergency impact range in the hazard handling plan and transmit it to the corresponding display terminal for display.

[0026] In one feasible approach

[0027] Step 3 includes:

[0028] Step 31: Draw several park access roads of the industrial park according to the park map of the industrial park, and map the range radius and danger level of each emergency impact range and each park access road into the large model to determine the access road safety level of different park access roads in each emergency impact range.

[0029] Step 32: Use visual technology to identify on-site personnel activity information in the industrial park, set corresponding priorities for each park passage based on the passage safety level, simulate the evacuation path corresponding to each on-site personnel in the large model, and mark several path turning points corresponding to each on-site personnel in the large model.

[0030] Step 33: Identify the hazard attributes corresponding to each of the aforementioned emergency impact ranges, determine the dangerous dwell time corresponding to each of the aforementioned path inflection points, use the dangerous dwell time to conduct a safety assessment of the number of people waiting to pass corresponding to the aforementioned path inflection points, and adjust the evacuation routes of the corresponding on-site personnel according to the assessment results;

[0031] Step 34: Determine the guided evacuation routes and dangerous prohibited routes of the industrial park based on the safety level of each of the park's passages, and construct the evacuation and disposal plan for the industrial park by combining the effective evacuation routes corresponding to each of the on-site personnel, and transmit it to the corresponding display terminal for display.

[0032] In one feasible approach

[0033] Also includes:

[0034] Collect inquiries from relevant users via terminals, answer the inquiries according to the hazard response plan and the evacuation plan, and provide feedback through the corresponding terminals.

[0035] In one feasible approach

[0036] Step 4 includes:

[0037] Step 41: When an emergency occurs in the industrial park, collect on-site emergency data of the industrial park in different dimensions, and reconstruct several real-time emergency measures of the industrial park based on the on-site emergency data;

[0038] Step 42: Determine the real-time emergency effect of the industrial park based on the emergency location corresponding to each of the real-time emergency measures, and identify the handling defects when implementing the real-time emergency measures;

[0039] Step 43: Construct on-site handling measures for the handling defects based on the hazardous disposal plan and the evacuation disposal plan;

[0040] Step 44: Based on the on-site handling measures, construct the corresponding emergency command password and transmit it to the target terminal closest to the handling defect for on-site command.

[0041] In one feasible approach

[0042] Also includes:

[0043] The safety hazards within the industrial park are statistically analyzed, and a hazard report for the industrial park is constructed and displayed.

[0044] This invention provides an emergency command and dispatch platform for industrial parks based on a large model, comprising:

[0045] The hazard identification module is used to identify safety hazards in industrial parks based on visual technology and to construct hazard mitigation plans for the industrial parks using a large model.

[0046] The hazard handling module is used to divide the industrial park into several emergency impact areas using the large model when an emergency occurs in the industrial park, and to construct a corresponding hazard handling plan for each emergency impact area.

[0047] The evacuation and disposal module is used to collect information on the number of people stranded within each of the aforementioned emergency impact areas, and to construct an evacuation and disposal plan for the on-site personnel based on the hazard attributes of the corresponding emergency impact areas.

[0048] The on-site emergency module is used to determine the deficiencies in the current emergency response based on the on-site emergency data of the industrial park, and generate corresponding emergency command commands to conduct on-site command.

[0049] In one feasible approach

[0050] The hazard handling module includes:

[0051] The video processing unit is used to monitor the real-time surveillance video of the industrial park in real time. When an emergency occurs in the industrial park, the real-time surveillance video is processed by frame segmentation, and the time point corresponding to each video frame is determined. Each video frame is converted into numerical data and combined with the corresponding time point to obtain a multi-dimensional array corresponding to each video frame.

[0052] The scope division unit is used to input each of the multidimensional arrays into the large model for data analysis based on time sequence, divide the industrial park into several emergency impact ranges according to the emergency danger locations and emergency danger attributes in the industrial park, and determine the range radius and danger level corresponding to each emergency impact range.

[0053] The training and analysis unit is used to deduce several affected features corresponding to each of the emergency impact ranges based on the emergency danger attributes, and to use a preset collaborative neural network to perform multi-angle learning and training on each of the affected features to obtain several scene parameters corresponding to each of the affected features. The unit analyzes the parameter logic between different scene parameters, determines the feature relationship between different affected features, and constructs a corresponding feature tree.

[0054] The feature analysis unit is used to construct the associated influence factor corresponding to each of the affected features based on the feature tree, construct a corresponding first feature treatment scheme for the target affected feature with the largest associated influence factor using the large model, determine the causal treatment result of the first feature treatment scheme for each of the affected features based on the feature tree, and construct a second treatment scheme for the corresponding affected feature based on the causal treatment result.

[0055] The solution generation unit is used to obtain the disposal solution corresponding to each of the affected features, construct the hazard disposal plan of the industrial park, find the hazard disposal solution corresponding to each of the emergency impact ranges in the hazard disposal plan, and transmit it to the corresponding display terminal for display.

[0056] The beneficial effects of the above technical solution are as follows: To reduce the probability of accidents in industrial parks, visual technology is used to identify safety hazards in industrial parks under normal circumstances, and the convenience of large-scale models is used to construct hazard mitigation plans for industrial parks. When an emergency occurs in an industrial park, the park is quickly divided into domino zones centered on the location of the emergency: a death zone, a seriously injured zone, and a slightly injured zone. Different hazard mitigation plans are then developed for different zones, effectively preventing the accident from escalating and minimizing losses within the industrial park. Simultaneously, the location of stranded personnel within each zone is determined, and corresponding evacuation plans are assigned to them. During emergency response, on-site first aid data is used to deduce current shortcomings and promptly grasp the progress of accident handling, thereby generating corresponding emergency command commands for on-site command and effective guidance of the handling work until the accident is eliminated. In this way, safety hazards and emergency situations within industrial parks can be identified in a short time, and corresponding measures can be taken to minimize the damage to the industrial park and maximize the safety of personnel within the park. Furthermore, the monitoring intensity and direction of the industrial park can be updated as technology advances, achieving the goal of continuous optimization and upgrading.

[0057] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0058] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0059] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0060] Figure 1 This is a schematic diagram illustrating the workflow of an emergency command and dispatch method for industrial parks based on a large model, as described in an embodiment of the present invention.

[0061] Figure 2 This is a schematic diagram of the composition of an emergency command and dispatch platform for industrial parks based on a large model, as described in an embodiment of the present invention. Detailed Implementation

[0062] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0063] Example 1

[0064] This embodiment provides an emergency command and dispatch method for industrial parks based on a large model, such as... Figure 1 Shown, including:

[0065] Step 1: Identify safety hazards in the industrial park based on visual technology, and construct a hazard mitigation plan for the industrial park using a large model;

[0066] Step 2: When an emergency occurs in the industrial park, the large model is used to divide the industrial park into several emergency impact areas, and a corresponding hazard disposal plan is constructed for each emergency impact area.

[0067] Step 3: Collect information on the number of people stranded within each of the aforementioned emergency impact areas, and construct an evacuation and disposal plan for the on-site personnel based on the hazard attributes of the corresponding emergency impact areas;

[0068] Step 4: Based on the on-site emergency data of the industrial park, determine the deficiencies in the current emergency response and generate corresponding emergency command commands to conduct on-site command.

[0069] In this example, a safety hazard represents a potential risk;

[0070] In this example, the hazard mitigation plan represents the method used to eliminate safety hazards in the industrial park;

[0071] In this example, the emergency situation refers to a sudden danger within the industrial park;

[0072] In this example, the emergency impact range includes: the area represented by the domino radius, the area represented by the death radius, the area represented by the serious injury radius, and the area represented by the minor injury radius;

[0073] In this example, the hazard response plan includes emergency repair and rescue, leak containment and environmental cleanup, and automatically matches emergency supplies such as ventilators and chemical protective suits in the park, and can realize intelligent scheduling of emergency supplies;

[0074] In this example, the evacuation plan included evacuation routes and recommendations.

[0075] In this example, the handling defect refers to an operational defect that occurs during on-site emergency response;

[0076] In this example, the emergency command code means that the emergency command code is played on-site via voice.

[0077] The working principle and beneficial effects of the above technical solution are as follows: To reduce the probability of accidents in industrial parks, visual technology is used to identify safety hazards in industrial parks under normal circumstances. The convenience of large-scale models is used to construct hazard mitigation plans for industrial parks. In the event of an emergency, the industrial park is quickly divided into domino zones centered on the location of the emergency: a death zone, a seriously injured zone, and a slightly injured zone. Different hazard mitigation plans are then developed for each zone, effectively preventing the accident from escalating and minimizing losses within the industrial park. Simultaneously, the location of stranded personnel within each zone is determined, and corresponding evacuation plans are assigned to them. During emergency response, on-site first aid data is used to deduce current shortcomings and promptly grasp the progress of accident handling, thereby generating corresponding emergency command commands for on-site command and effective guidance of the handling work until the accident is eliminated. This method allows for the rapid identification of safety hazards and emergency situations within industrial parks, enabling appropriate handling and minimizing damage to the industrial park while maximizing the safety of personnel within the park. Furthermore, the monitoring intensity and direction of the industrial park can be updated as technology advances, achieving continuous optimization and upgrading.

[0078] Example 2

[0079] Based on Example 1, the emergency command and dispatch method for industrial parks based on a large model, step 1 includes:

[0080] Step 11: Use visual recognition technology to extract features from the historical surveillance video of the industrial park to obtain several historical local features of the industrial park. Simultaneously compare the historical local features corresponding to the same area of ​​the park to obtain the local feature period corresponding to each area of ​​the park and construct the coarse feature change trend of each area of ​​the park.

[0081] Step 12: Use visual recognition technology to extract features from the real-time monitoring video of the industrial park to obtain several real-time local features of the industrial park. Use the coarse feature transformation trend to compare the corresponding real-time local features and extract the target real-time local features that are inconsistent with the coarse feature transformation trend.

[0082] Step 13: Locate the target park area corresponding to the real-time local features of the target in the real-time monitoring video, deduce the project parameters of the target park area based on the regional engineering projects of the target park area and the real-time local features of the target, and determine the safety hazards of the target park area;

[0083] Step 14: Input the safety hazards and the regional engineering projects into the large model to trace the source of the hazards, obtain the causes and locations of the hazards in the target park area, screen the corresponding initial hazard handling plans, decompose the regional engineering projects, determine the engineering parameters corresponding to the location of the hazards, and use the engineering parameters to adjust the data of the initial hazard handling plans to obtain the hazard handling plan for the industrial park.

[0084] In this example, historical local features represent the characteristics of each area within the industrial park;

[0085] In this example, the local feature periodicity represents the periodic characteristics exhibited by the park area;

[0086] In this example, the coarse feature transformation trend represents the trend of historical local features in a park area as a result of periodic transformation;

[0087] In this example, the initial hazard mitigation plan is a general approach, but it is not as specific as the hazard mitigation plan.

[0088] The working principle and beneficial effects of the above technical solution are as follows: First, visual recognition technology is used to analyze the features of historical surveillance videos to determine the coarse feature transformation trends of the park area. Then, visual recognition technology is used again to analyze real-time surveillance videos to determine the relationship between each real-time local feature in the industrial park and the corresponding coarse feature transformation trend. This allows for the extraction of target real-time local features that are inconsistent with the coarse feature transformation trend, further locating the target park area corresponding to the feature. Then, based on the regional engineering projects and real-time local features of the area, the project parameters are derived to determine the safety hazards in the target park area. Finally, the cause and location of the hazards are determined through hazard tracing, and corresponding initial hazard mitigation plans are extracted and adjusted to construct a hazard mitigation plan for the industrial park. This approach takes into account the changing conditions within the industrial park for hazard identification and the construction of corresponding plans, effectively reducing identification errors and improving the safety of the industrial park.

[0089] Example 3

[0090] Based on Example 2, the industrial park emergency command and dispatch method based on a large model further includes:

[0091] Using the real-time local features, the regional engineering progress corresponding to the park area is determined, and the safety hazard database of the industrial park is updated according to the regional engineering progress.

[0092] The working principle and beneficial effects of the above technical solution are as follows: Since industrial parks are prone to different safety hazards under different scenarios, the safety hazard database is updated by inferring the regional engineering progress of the park area based on real-time local features, thereby improving the accuracy of safety hazard identification.

[0093] Example 4

[0094] Based on Example 1, the emergency command and dispatch method for industrial parks based on a large model, step 2 includes:

[0095] Step 21: Monitor the real-time video of the industrial park in real time. When an emergency occurs in the industrial park, perform frame-by-frame processing on the real-time video and determine the time point corresponding to each video frame. Convert each video frame into numerical data and combine the corresponding time points to obtain a multi-dimensional array corresponding to each video frame.

[0096] Step 22: Input each of the multidimensional arrays into the large model for data analysis based on the time sequence. Divide the industrial park into several emergency impact ranges according to the emergency danger locations and emergency danger attributes in the industrial park, and determine the range radius and danger level corresponding to each emergency impact range.

[0097] Step 23: Based on the emergency danger attributes, deduce several affected features corresponding to each emergency impact range, use a preset collaborative neural network to perform multi-angle learning and training on each affected feature, obtain several scene parameters corresponding to each affected feature, analyze the parameter logic between different scene parameters, determine the feature relationship between different affected features, and construct the corresponding feature tree.

[0098] Step 24: Construct the associated influence factor corresponding to each of the affected features based on the feature tree; use the large model to construct a corresponding first feature treatment scheme for the target affected feature with the largest associated influence factor; determine the causal treatment result of the first feature treatment scheme for each of the affected features based on the feature tree; and construct a second treatment scheme for the corresponding affected feature based on the causal treatment result.

[0099] Step 25: Obtain the handling plan corresponding to each of the affected features and construct the hazard handling plan for the industrial park. Find the hazard handling plan corresponding to each emergency impact range in the hazard handling plan and transmit it to the corresponding display terminal for display.

[0100] In this example, frame splitting refers to the process of dividing real-time monitoring video into image frames;

[0101] In this example, one video frame corresponds to one time point, which represents the moment when the video frame was generated;

[0102] In this example, numerical data represents the result of using data to express video frames;

[0103] In this example, the multidimensional array represents the result of using the three dimensions of RGB to express the video frame;

[0104] In this example, the affected feature represents the feature that appears when an emergency impact range is affected by the emergency danger attribute.

[0105] In this example, the pre-defined collaborative neural network is represented as a network used for multi-time learning;

[0106] In this example, multi-angle learning training represents the process of analyzing the manifestation of the affected features in different scenarios;

[0107] In this example, the scene parameters represent the parameters of the affected feature under different scenes;

[0108] In this example, the parameter logic indicates that there is a logical relationship between the scene parameters of different affected features;

[0109] In this example, the feature tree represents a binary tree used to express the logical relationships between different affected features;

[0110] In this example, the correlation impact factor represents the degree to which an affected feature is interfered with by other affected features;

[0111] In this example, the causal treatment result represents the impact of the first feature treatment scheme on the affected feature.

[0112] The working principle and beneficial effects of the above technical solution are as follows: While the projects within an industrial park are complex, the underlying logic is clear. When an emergency occurs, it's not sufficient to only address the location of the incident; the situation of related departments must also be considered. Therefore, in the event of an emergency, the real-time monitoring video is first processed by frame segmentation and numerical conversion, resulting in several sets of multi-dimensional arrays arranged in chronological order. A large model is then used to analyze these arrays, dividing the industrial park into several emergency impact zones. The affected characteristics of each emergency impact zone when disturbed by an emergency are derived. These affected characteristics are then trained from multiple perspectives and applied to different scenarios. The feature tree is constructed using the parameter logic between scene parameters, which determines the associated influencing factors of each affected feature. In order to quickly build a response plan, the target affected feature with the largest associated influencing factor is selected first to build the corresponding first response plan. The synchronous response results of this plan for the remaining affected features are determined. Then, associated influencing factors with progressively smaller associated influencing factors are selected and corresponding response plans are built. Finally, a hazard response plan for the industrial park is constructed, and a hazard response plan for each emergency impact range is determined. In this way, not only can emergencies in the industrial park be dealt with in a timely manner, but the normal operation of other areas in the industrial park can also be guaranteed, and losses can be reduced.

[0113] Example 5

[0114] Based on Example 1, the emergency command and dispatch method for industrial parks based on a large model, step 3 includes:

[0115] Step 31: Draw several park access roads of the industrial park according to the park map of the industrial park, and map the range radius and danger level of each emergency impact range and each park access road into the large model to determine the access road safety level of different park access roads in each emergency impact range.

[0116] Step 32: Use visual technology to identify on-site personnel activity information in the industrial park, set corresponding priorities for each park passage based on the passage safety level, simulate the evacuation path corresponding to each on-site personnel in the large model, and mark several path turning points corresponding to each on-site personnel in the large model.

[0117] Step 33: Identify the hazard attributes corresponding to each of the aforementioned emergency impact ranges, determine the dangerous dwell time corresponding to each of the aforementioned path inflection points, use the dangerous dwell time to conduct a safety assessment of the number of people waiting to pass corresponding to the aforementioned path inflection points, and adjust the evacuation routes of the corresponding on-site personnel according to the assessment results;

[0118] Step 34: Determine the guided evacuation routes and dangerous prohibited routes of the industrial park based on the safety level of each of the park's passages, and construct the evacuation and disposal plan for the industrial park by combining the effective evacuation routes corresponding to each of the on-site personnel, and transmit it to the corresponding display terminal for display.

[0119] In this example, the guided evacuation routes indicate the evacuation routes that on-site personnel can choose according to actual needs, while the dangerous prohibited routes indicate routes that are prone to serious accidents and are not recommended for on-site personnel to enter.

[0120] The working principle and beneficial effects of the above technical solution are as follows: When an emergency occurs in the industrial park, the first priority is to ensure the personal safety of personnel within the park to the greatest extent possible. First, the park's access routes are drawn based on the park map. The safety level of each access route is determined by combining the radius and hazard level of each emergency impact area, thus setting corresponding priorities for the access routes. Then, by locating the positions of on-site personnel and continuously adjusting them, an evacuation route is created for each on-site person. Simultaneously, based on the access route safety level, guided evacuation routes and dangerous stationary routes within the park are determined to provide evacuation guidance to on-site personnel. Finally, an evacuation and disposal plan is constructed for reference by on-site personnel and off-site management personnel to guide on-site personnel to evacuate as quickly as possible and ensure their personal safety.

[0121] Example 6

[0122] Based on Example 1, the aforementioned emergency command and dispatch method for industrial parks based on a large model further includes:

[0123] Collect inquiries from relevant users via terminals, answer the inquiries according to the hazard response plan and the evacuation plan, and provide feedback through the corresponding terminals.

[0124] The working principle and beneficial effects of the above technical solution are as follows: AI question answering can guide on-site personnel to make the correct response in a short period of time.

[0125] Example 7

[0126] Based on Example 1, the emergency command and dispatch method for industrial parks based on a large model, step 4 includes:

[0127] Step 41: When an emergency occurs in the industrial park, collect on-site emergency data of the industrial park in different dimensions, and reconstruct several real-time emergency measures of the industrial park based on the on-site emergency data;

[0128] Step 42: Determine the real-time emergency effect of the industrial park based on the emergency location corresponding to each of the real-time emergency measures, and identify the handling defects when implementing the real-time emergency measures;

[0129] Step 43: Construct on-site handling measures for the handling defects based on the hazardous disposal plan and the evacuation disposal plan;

[0130] Step 44: Based on the on-site handling measures, construct the corresponding emergency command password and transmit it to the target terminal closest to the handling defect for on-site command.

[0131] The working principle and beneficial effects of the above technical solution are as follows: By analyzing the handling defects that occur during the emergency, the solution can provide on-site personnel with commands. This can not only deal with the sudden situation on-site, but also provide on-site personnel with technical references, guide them to deal with the accident on-site as soon as possible, and leave the site as soon as possible, thus ensuring the safety of the industrial park and on-site personnel.

[0132] Example 8

[0133] Based on Example 2, the industrial park emergency command and dispatch method based on a large model further includes:

[0134] The safety hazards within the industrial park are statistically analyzed, and a hazard report for the industrial park is constructed and displayed.

[0135] Example 9

[0136] This embodiment provides an emergency command and dispatch platform for industrial parks based on a large model, such as... Figure 2 Shown, including:

[0137] The hazard identification module is used to identify safety hazards in industrial parks based on visual technology and to construct hazard mitigation plans for the industrial parks using a large model.

[0138] The hazard handling module is used to divide the industrial park into several emergency impact areas using the large model when an emergency occurs in the industrial park, and to construct a corresponding hazard handling plan for each emergency impact area.

[0139] The evacuation and disposal module is used to collect information on the number of people stranded within each of the aforementioned emergency impact areas, and to construct an evacuation and disposal plan for the on-site personnel based on the hazard attributes of the corresponding emergency impact areas.

[0140] The on-site emergency module is used to determine the deficiencies in the current emergency response based on the on-site emergency data of the industrial park, and generate corresponding emergency command commands to conduct on-site command.

[0141] In this example, a safety hazard represents a potential risk;

[0142] In this example, the hazard mitigation plan represents the method used to eliminate safety hazards in the industrial park;

[0143] In this example, the emergency situation refers to a sudden danger within the industrial park;

[0144] In this example, the emergency impact range includes: the area represented by the domino radius, the area represented by the death radius, the area represented by the serious injury radius, and the area represented by the minor injury radius;

[0145] In this example, the hazard response plan includes emergency repair and rescue, leak containment and environmental cleanup, and automatically matches emergency supplies such as ventilators and chemical protective suits in the park, and can realize intelligent scheduling of emergency supplies;

[0146] In this example, the evacuation plan included evacuation routes and recommendations.

[0147] In this example, the handling defect refers to an operational defect that occurs during on-site emergency response;

[0148] In this example, the emergency command code means that the emergency command code is played on-site via voice.

[0149] The working principle and beneficial effects of the above technical solution are as follows: To reduce the probability of accidents in industrial parks, visual technology is used to identify safety hazards in industrial parks under normal circumstances. The convenience of large-scale models is used to construct hazard mitigation plans for industrial parks. In the event of an emergency, the industrial park is quickly divided into domino zones centered on the location of the emergency: a death zone, a seriously injured zone, and a slightly injured zone. Different hazard mitigation plans are then developed for each zone, effectively preventing the accident from escalating and minimizing losses within the industrial park. Simultaneously, the location of stranded personnel within each zone is determined, and corresponding evacuation plans are assigned to them. During emergency response, on-site first aid data is used to deduce current shortcomings and promptly grasp the progress of accident handling, thereby generating corresponding emergency command commands for on-site command and effective guidance of the handling work until the accident is eliminated. This method allows for the rapid identification of safety hazards and emergency situations within industrial parks, enabling appropriate handling and minimizing damage to the industrial park while maximizing the safety of personnel within the park. Furthermore, the monitoring intensity and direction of the industrial park can be updated as technology advances, achieving continuous optimization and upgrading.

[0150] Example 10

[0151] Based on Example 9, the emergency command and dispatch platform for industrial parks based on a large model, wherein the hazard handling module includes:

[0152] The video processing unit is used to monitor the real-time surveillance video of the industrial park in real time. When an emergency occurs in the industrial park, the real-time surveillance video is processed by frame segmentation, and the time point corresponding to each video frame is determined. Each video frame is converted into numerical data and combined with the corresponding time point to obtain a multi-dimensional array corresponding to each video frame.

[0153] The scope division unit is used to input each of the multidimensional arrays into the large model for data analysis based on time sequence, divide the industrial park into several emergency impact ranges according to the emergency danger locations and emergency danger attributes in the industrial park, and determine the range radius and danger level corresponding to each emergency impact range.

[0154] The training and analysis unit is used to deduce several affected features corresponding to each of the emergency impact ranges based on the emergency danger attributes, and to use a preset collaborative neural network to perform multi-angle learning and training on each of the affected features to obtain several scene parameters corresponding to each of the affected features. The unit analyzes the parameter logic between different scene parameters, determines the feature relationship between different affected features, and constructs a corresponding feature tree.

[0155] The feature analysis unit is used to construct the associated influence factor corresponding to each of the affected features based on the feature tree, construct a corresponding first feature treatment scheme for the target affected feature with the largest associated influence factor using the large model, determine the causal treatment result of the first feature treatment scheme for each of the affected features based on the feature tree, and construct a second treatment scheme for the corresponding affected feature based on the causal treatment result.

[0156] The solution generation unit is used to obtain the disposal solution corresponding to each of the affected features, construct the hazard disposal plan of the industrial park, find the hazard disposal solution corresponding to each of the emergency impact ranges in the hazard disposal plan, and transmit it to the corresponding display terminal for display.

[0157] In this example, frame splitting refers to the process of dividing real-time monitoring video into image frames;

[0158] In this example, one video frame corresponds to one time point, which represents the moment when the video frame was generated;

[0159] In this example, numerical data represents the result of using data to express video frames;

[0160] In this example, the multidimensional array represents the result of using the three dimensions of RGB to express the video frame;

[0161] In this example, the affected feature represents the feature that appears when an emergency impact range is affected by the emergency danger attribute.

[0162] In this example, the pre-defined collaborative neural network is represented as a network used for multi-time learning;

[0163] In this example, multi-angle learning training represents the process of analyzing the manifestation of the affected features in different scenarios;

[0164] In this example, the scene parameters represent the parameters of the affected feature under different scenes;

[0165] In this example, the parameter logic indicates that there is a logical relationship between the scene parameters of different affected features;

[0166] In this example, the feature tree represents a binary tree used to express the logical relationships between different affected features;

[0167] In this example, the correlation impact factor represents the degree to which an affected feature is interfered with by other affected features;

[0168] In this example, the causal treatment result represents the impact of the first feature treatment scheme on the affected feature.

[0169] The working principle and beneficial effects of the above technical solution are as follows: While the projects within an industrial park are complex, the underlying logic is clear. When an emergency occurs, it's not sufficient to only address the location of the incident; the situation of related departments must also be considered. Therefore, in the event of an emergency, the real-time monitoring video is first processed by frame segmentation and numerical conversion, resulting in several sets of multi-dimensional arrays arranged in chronological order. A large model is then used to analyze these arrays, dividing the industrial park into several emergency impact zones. The affected characteristics of each emergency impact zone when disturbed by an emergency are derived. These affected characteristics are then trained from multiple perspectives and applied to different scenarios. The feature tree is constructed using the parameter logic between scene parameters, which determines the associated influencing factors of each affected feature. In order to quickly build a response plan, the target affected feature with the largest associated influencing factor is selected first to build the corresponding first response plan. The synchronous response results of this plan for the remaining affected features are determined. Then, associated influencing factors with progressively smaller associated influencing factors are selected and corresponding response plans are built. Finally, a hazard response plan for the industrial park is constructed, and a hazard response plan for each emergency impact range is determined. In this way, not only can emergencies in the industrial park be dealt with in a timely manner, but the normal operation of other areas in the industrial park can also be guaranteed, and losses can be reduced.

[0170] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for emergency command and dispatch in industrial parks based on a large model, characterized in that, include: Step 1: Use visual recognition technology to extract features from the historical surveillance video of the industrial park to obtain several historical local features of the industrial park. Simultaneously compare the historical local features corresponding to the same area of ​​the park to obtain the local feature period corresponding to each area of ​​the park and construct the coarse feature change trend of each area of ​​the park. Visual recognition technology is used to extract features from the real-time monitoring video of the industrial park to obtain several real-time local features of the industrial park. The coarse feature change trend is used to compare the corresponding real-time local features and extract the target real-time local features that are inconsistent with the coarse feature change trend. In the real-time monitoring video, locate the target park area corresponding to the real-time local features of the target, deduce the project parameters of the target park area based on the regional engineering projects of the target park area and the real-time local features of the target, and determine the safety hazards of the target park area; The safety hazards and the regional engineering projects are input into the large model to trace the source of the hazards, obtain the causes and locations of the hazards in the target park area, screen the corresponding initial hazard handling plans, decompose the regional engineering projects, determine the engineering parameters corresponding to the location of the hazards, and use the engineering parameters to adjust the data of the initial hazard handling plans to obtain the hazard handling plan for the industrial park. Step 2: Monitor the real-time video of the industrial park in real time. When an emergency occurs in the industrial park, the real-time video is processed into frames and the time point corresponding to each video frame is determined. Each video frame is converted into numerical data and combined with the corresponding time point to obtain a multi-dimensional array corresponding to each video frame. Based on the time sequence, each of the multidimensional arrays is input into the large model for data analysis. According to the emergency danger location and emergency danger attributes in the industrial park, the industrial park is divided into several emergency impact ranges, and the range radius and danger level corresponding to each emergency impact range are determined. Based on the emergency danger attributes, several affected features corresponding to each emergency impact range are derived. A preset collaborative neural network is used to perform multi-angle learning and training on each affected feature to obtain several scene parameters corresponding to each affected feature. The parameter logic between different scene parameters is analyzed to determine the feature relationship between different affected features and to construct a corresponding feature tree. Based on the feature tree, construct the associated influence factor corresponding to each of the affected features. Using the large model, construct a corresponding first feature treatment scheme for the target affected feature with the largest associated influence factor. Based on the feature tree, determine the causal treatment result of the first feature treatment scheme for each of the affected features. Based on the causal treatment result, construct a second treatment scheme for the corresponding affected feature. A hazard response plan for the industrial park is constructed by obtaining the corresponding response plan for each affected feature. The hazard response plan is then used to find the hazard response plan corresponding to each emergency impact range and transmitted to the corresponding display terminal for display. Step 3: Collect information on the number of people stranded within each of the aforementioned emergency impact areas, and construct an evacuation and disposal plan for the on-site personnel based on the hazard attributes of the corresponding emergency impact areas; Step 4: Based on the on-site emergency data of the industrial park, determine the deficiencies in the current emergency response and generate corresponding emergency command commands to conduct on-site command.

2. The emergency command and dispatch method for industrial parks based on a large model as described in claim 1, characterized in that, Also includes: Using the real-time local features, the regional engineering progress corresponding to the park area is determined, and the safety hazard database of the industrial park is updated according to the regional engineering progress.

3. The emergency command and dispatch method for industrial parks based on a large model as described in claim 1, characterized in that, Step 3 includes: Step 31: Draw several park access roads of the industrial park according to the park map of the industrial park, and map the range radius and danger level of each emergency impact range and each park access road into the large model to determine the access road safety level of different park access roads in each emergency impact range. Step 32: Use visual technology to identify on-site personnel activity information in the industrial park, set corresponding priorities for each park passage based on the passage safety level, simulate the evacuation path corresponding to each on-site personnel in the large model, and mark several path turning points corresponding to each on-site personnel in the large model. Step 33: Identify the hazard attributes corresponding to each of the aforementioned emergency impact ranges, determine the dangerous dwell time corresponding to each of the aforementioned path inflection points, use the dangerous dwell time to conduct a safety assessment of the personnel waiting to pass through the corresponding path inflection points, and adjust the evacuation routes of the corresponding on-site personnel according to the assessment results; Step 34: Determine the guided evacuation routes and dangerous prohibited routes of the industrial park based on the safety level of each of the park's passages, and construct the evacuation and disposal plan for the industrial park by combining the effective evacuation routes corresponding to each of the on-site personnel, and transmit it to the corresponding display terminal for display.

4. The emergency command and dispatch method for industrial parks based on a large model as described in claim 1, characterized in that, Also includes: Collect inquiries from relevant users via terminals, answer the inquiries according to the hazard response plan and the evacuation plan, and provide feedback through the corresponding terminals.

5. The emergency command and dispatch method for industrial parks based on a large model as described in claim 1, characterized in that, Step 4 includes: Step 41: When an emergency occurs in the industrial park, collect on-site emergency data of the industrial park in different dimensions, and reconstruct several real-time emergency measures of the industrial park based on the on-site emergency data; Step 42: Determine the real-time emergency effect of the industrial park based on the emergency location corresponding to each of the real-time emergency measures, and identify the handling defects when implementing the real-time emergency measures; Step 43: Construct on-site handling measures for the handling defects based on the hazardous disposal plan and the evacuation disposal plan; Step 44: Based on the on-site handling measures, construct the corresponding emergency command password and transmit it to the target terminal closest to the handling defect for on-site command.

6. The emergency command and dispatch method for industrial parks based on a large model as described in claim 1, characterized in that, Also includes: The safety hazards within the industrial park are statistically analyzed, and a hazard report for the industrial park is constructed and displayed.

7. An emergency command and dispatch platform for industrial parks based on a large model, characterized in that, include: The hazard identification module is used to extract features from the historical surveillance videos of the industrial park using visual recognition technology, obtain several historical local features of the industrial park, and synchronously compare the historical local features corresponding to the same area of ​​the park to obtain the local feature period corresponding to each area of ​​the park, and construct the coarse feature change trend of each area of ​​the park. Visual recognition technology is used to extract features from the real-time monitoring video of the industrial park to obtain several real-time local features of the industrial park. The coarse feature change trend is used to compare the corresponding real-time local features and extract the target real-time local features that are inconsistent with the coarse feature change trend. In the real-time monitoring video, locate the target park area corresponding to the real-time local features of the target, deduce the project parameters of the target park area based on the regional engineering projects of the target park area and the real-time local features of the target, and determine the safety hazards of the target park area; The safety hazards and the regional engineering projects are input into the large model to trace the source of the hazards, obtain the causes and locations of the hazards in the target park area, screen the corresponding initial hazard handling plans, decompose the regional engineering projects, determine the engineering parameters corresponding to the location of the hazards, and use the engineering parameters to adjust the data of the initial hazard handling plans to obtain the hazard handling plan for the industrial park. The hazard handling module is used to monitor the real-time video of the industrial park in real time. When an emergency occurs in the industrial park, the real-time video is processed into frames, and the time point corresponding to each video frame is determined. Each video frame is converted into numerical data, and a multi-dimensional array corresponding to each video frame is obtained by combining the corresponding time point. Based on the time sequence, each of the multidimensional arrays is input into the large model for data analysis. According to the emergency danger location and emergency danger attributes in the industrial park, the industrial park is divided into several emergency impact ranges, and the range radius and danger level corresponding to each emergency impact range are determined. Based on the emergency danger attributes, several affected features corresponding to each emergency impact range are derived. A preset collaborative neural network is used to perform multi-angle learning and training on each affected feature to obtain several scene parameters corresponding to each affected feature. The parameter logic between different scene parameters is analyzed to determine the feature relationship between different affected features and to construct a corresponding feature tree. Based on the feature tree, construct the associated influence factor corresponding to each of the affected features. Using the large model, construct a corresponding first feature treatment scheme for the target affected feature with the largest associated influence factor. Based on the feature tree, determine the causal treatment result of the first feature treatment scheme for each of the affected features. Based on the causal treatment result, construct a second treatment scheme for the corresponding affected feature. A hazard response plan for the industrial park is constructed by obtaining the corresponding response plan for each affected feature. The hazard response plan is then used to find the hazard response plan corresponding to each emergency impact range and transmitted to the corresponding display terminal for display. The evacuation and disposal module is used to collect information on the number of people stranded within each of the aforementioned emergency impact areas, and to construct an evacuation and disposal plan for the on-site personnel based on the hazard attributes of the corresponding emergency impact areas. The on-site emergency module is used to determine the deficiencies in the current emergency response based on the on-site emergency data of the industrial park, and generate corresponding emergency command commands to conduct on-site command.

Images