Chemical industrial park safety assessment method and system combined with video diagnostic tool

Abstract

The invention discloses a chemical industry park safety assessment method and system combined with a video diagnosis tool, and relates to the technical field of safety management, and the method comprises the steps: determining a plurality of video monitoring regions of a target chemical industry park and corresponding preset abnormal behavior sets; defining risk accompanying features and shielding features based on the abnormal behavior set, and constructing a primary video frame screening tool; performing preliminary screening on the video acquisition results by using a screening tool, and generating a screening frame set carrying a detection behavior identifier; judging whether the processing capacity of the screening frame set exceeds a preset concurrency threshold value or not; if yes, risk real-time analysis is executed according to the detection behavior identification, real-time and delay safety analysis is carried out according to the risk priority, and a regional safety analysis result is output. Therefore, the technical effects of improving the monitoring data processing efficiency and enhancing the security identification accuracy and responsiveness are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of safety management, and particularly to a safety assessment method and system for chemical industrial parks combined with video diagnostic tools. Background Art

[0002] At present, each enterprise has invested a large number of different types of video monitoring devices on site, aiming to solve the problems of timely detecting the behavior problems of on-site personnel and the unsafe states of objects, and making timely corrections to ensure on-site safety. However, with the continuous expansion of the scale of chemical industrial parks and the increasing complexity of production activities, these traditional methods gradually show deficiencies when facing a large amount of video data and diverse safety hazards.

[0003] With the continuous expansion of the scale of chemical industrial parks and the increasing complexity of production activities, the large amount of video footage and data volume make it impossible for personnel to view in a timely manner, and the frequently upgraded video AI algorithms often cause problems with historical data being unable to be effectively evaluated. Summary of the Invention

[0004] The present invention provides a safety assessment method and system for chemical industrial parks combined with video diagnostic tools to solve the technical problems in the prior art that the monitoring data volume is too large, resulting in difficult processing and inability to perform safety identification in a timely and accurate manner, and to achieve the technical effects of improving the processing efficiency of monitoring data, enhancing the accuracy and responsiveness of safety identification.

[0005] In a first aspect, the present invention provides a safety assessment method for chemical industrial parks combined with video diagnostic tools, wherein the safety assessment method for chemical industrial parks combined with video diagnostic tools includes: Determine multiple preset abnormal behavior sets for multiple video monitoring areas within the target chemical industrial park.

[0006] Based on the multiple preset abnormal behavior sets, define the risk accompanying features and shielding features in the video images, and construct multiple primary video frame screening tools with the definition results.

[0007] Use the multiple primary video frame screening tools to perform preliminary risk frame screening on the video acquisition results of the multiple video monitoring areas, generate multiple screened frame sets, and any screened frame carries a detection behavior identifier.

[0008] Analyze whether the data processing volume of the multiple screened frame sets is greater than a preset concurrent processing threshold.

[0009] If so, perform real-time risk analysis for each area based on the detection behavior identifier, and call the multiple primary video frame screening tools according to the real-time risk priority to perform real-time safety analysis and delayed safety analysis on the multiple screened frame sets, and generate multiple area safety analysis results.

[0010] In a feasible implementation manner, after generating multiple regional security analysis results, it further includes: Read a preset evaluation period.

[0011] Perform statistics on the security analysis results of multiple video monitoring regions with the preset evaluation period, and generate an unsafe behavior statistics table.

[0012] Based on the unsafe behavior statistics table, perform security evaluation on the multiple video monitoring regions, and generate multiple security evaluation results.

[0013] In a feasible implementation manner, any one of the multiple preset abnormal behavior sets includes unsafe behaviors of people and unsafe states of objects.

[0014] In a feasible implementation manner, based on the multiple preset abnormal behavior sets, define risk accompanying features and shielding features in video images, and construct multiple primary video frame screening tools with the definition results, including: For each abnormal behavior in the multiple preset abnormal behavior sets, obtain the warning shielding features defined by the user, and generate shielding feature definition results.

[0015] For each abnormal behavior in the multiple preset abnormal behavior sets, collect the dynamic difference features of the pictures in the warning trigger scenario and the non-trigger scenario, and generate risk accompanying feature definition results.

[0016] Construct a dynamic detection model with the risk accompanying feature definition results, and optimize it with the shielding feature definition results to generate the multiple primary video frame screening tools.

[0017] In a feasible implementation manner, based on the detection behavior identifier, perform real-time risk analysis on each region, and call the multiple primary video frame screening tools according to the real-time risk priority to perform real-time security analysis and delayed security analysis on multiple screening frame sets, and generate multiple regional security analysis results, including: Based on the detection behavior identifier, perform trigger risk evaluation on the unsafe behaviors in each region, and construct a real-time risk priority.

[0018] Combine the preset concurrency processing threshold and the real-time risk priority to match the real-time security analysis mode and the delayed security mode for the multiple screening frame sets, and generate real-time mode matching results and delayed mode matching results.

[0019] According to the real-time mode matching results and the delayed mode matching results, call the multiple primary video frame screening tools to perform security analysis on the multiple screening frame sets, and generate the multiple regional security analysis results.

[0020] In a feasible implementation manner, based on the detection behavior identifier, perform a triggering risk assessment of unsafe behaviors in each area, and construct a real-time risk priority, including: Based on the detection behavior identifier, perform a triggering feature fitting of unsafe behaviors in each area, and construct multiple triggering features.

[0021] Perform a triggering risk assessment with the multiple triggering features, including the risk of the behavior itself and the risk of linkage diffusion, and generate multiple risk evaluation values.

[0022] Arrange the multiple risk evaluation values in descending order to generate the real-time risk priority.

[0023] In a feasible implementation manner, the method for determining the preset concurrency processing threshold includes: Determine the deployment environment information of multiple primary video frame screening tools.

[0024] Perform a twin simulation with the deployment environment information to determine the maximum stable concurrency.

[0025] Generate the preset concurrency processing threshold with the maximum stable concurrency.

[0026] In a feasible implementation manner, the maximum stable concurrency includes the frame rate that simultaneously satisfies the preset bottleneck usage rate of the CPU / GPU and the preset unstable conditions.

[0027] In a feasible implementation manner, after generating multiple security assessment results, it further includes: Read the continuous security assessment results under the preset assessment period, and identify the high-frequency unsafe behaviors and the corresponding high-frequency unsafe areas where the continuous repetition times of the unsafe behaviors are greater than the preset times.

[0028] Retrieve the event occurrence videos corresponding to the high-frequency unsafe behaviors, compare them with the monitoring videos corresponding to the low-frequency unsafe behaviors with continuous repetition times less than the preset times, determine the unsafe factors, and send them to the management end for reminder.

[0029] In a second aspect, the present invention further provides a chemical industrial park safety assessment system combined with a video diagnosis tool. Among them, the chemical industrial park safety assessment system combined with the video diagnosis tool includes: A behavior preset module, used to determine multiple preset abnormal behavior sets in multiple video monitoring areas within the target chemical industrial park.

[0030] A screening tool construction module, used to define the risk accompanying features and shielding features in the video image based on the multiple preset abnormal behavior sets, and construct multiple primary video frame screening tools with the definition results.

[0031] A preliminary risk frame screening module is used to perform preliminary risk frame screening on the video acquisition results of the multiple video monitoring areas through the multiple primary video frame screening tools, generate multiple screened frame sets, and any screened frame carries a detection behavior identifier.

[0032] A processing volume analysis module is used to analyze whether the data processing volume of the multiple screened frame sets is greater than a preset concurrent processing threshold.

[0033] A security analysis module is used to, if so, perform real-time risk analysis on each area based on the detection behavior identifier, and call the multiple primary video frame screening tools according to the real-time risk priority to perform real-time security analysis and delayed security analysis on the multiple screened frame sets, and generate multiple area security analysis results.

[0034] The present invention discloses a chemical industrial park safety assessment method and system combined with video diagnosis tools, including: determining multiple preset abnormal behavior sets corresponding to multiple video monitoring areas in a target chemical industrial park; based on the multiple preset abnormal behavior sets, defining various risk accompanying features and shielding features, and constructing several primary video frame screening modules accordingly; using the multiple primary video frame screening modules to perform preliminary screening on the collected videos of each video monitoring area, extracting a risk frame set containing a detection behavior identifier, and forming multiple preliminary screened frame sets; analyzing the data processing volume required for the multiple screened frame sets above to determine whether it exceeds a preset concurrent processing capacity threshold; if it exceeds, based on the detection behavior identifier carried in the frame, perform real-time risk assessment of each monitoring area, sort the call priorities of the primary video frame screening modules according to the assessment results, perform real-time security analysis and delayed security analysis respectively, and output corresponding area security analysis results. The chemical industrial park safety assessment method and system combined with video diagnosis tools disclosed by the present invention solve the technical problems of difficult processing due to excessive monitoring data volume and inability to perform safety identification in a timely and accurate manner, and achieve the technical effects of improving the monitoring data processing efficiency, enhancing the accuracy and responsiveness of safety identification. Description of the Drawings

[0035] Figure 1 It is a schematic flow chart of the chemical industrial park safety assessment method combined with video diagnosis tools of the present invention.

[0036] Figure 2 It is a schematic structural diagram of the chemical industrial park safety assessment system combined with video diagnosis tools of the present invention.

[0037] Description of the reference numerals: Behavior preset module 11, screening tool construction module 12, preliminary risk frame screening module 13, processing volume analysis module 14, security analysis module 15. Detailed Embodiments

[0038] The above technical solutions will be described in detail below in conjunction with the specification drawings and specific embodiments to better understand the above technical solutions. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments that only explain the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention. In addition, it should be noted that for the sake of convenience of description, only parts related to the present invention are shown in the drawings, rather than all of them.

[0039] Example 1, as Figure 1 is a schematic flow chart of a chemical industrial park safety assessment method combining video diagnosis tools. Among them, the chemical industrial park safety assessment method combining video diagnosis tools includes: S100: Determine multiple preset abnormal behavior sets for multiple video monitoring areas in the target chemical industrial park.

[0040] Specifically, first, according to the specific chemical production links and types included in the target chemical industrial park, determine the corresponding abnormal behaviors and output them as multiple preset abnormal behavior sets. The multiple preset abnormal behavior sets can correspond to multiple video monitoring scopes divided according to different safety requirements in the chemical industrial park, such as production workshops, storage areas, around reaction kettles, etc. Among them, the preset abnormal behavior sets are determined by professional technical personnel based on the production process, historical accident data, and safety specification requirements of the chemical industrial park.

[0041] In some embodiments, any one of the multiple preset abnormal behavior sets includes unsafe behaviors of people and unsafe states of objects.

[0042] Specifically, the preset abnormal behavior set refers to a series of templates of possible unsafe behaviors set in advance according to the characteristics of different areas in the chemical industrial park, including unsafe behaviors of people (such as illegal operations, not wearing safety helmets, etc.) and unsafe states of objects (such as equipment leaks, irregular material stacking, etc.).

[0043] The above process provides the basis and direction for subsequent video diagnosis and analysis, ensuring that the video diagnosis tool can specifically identify and screen potential safety risks. For example, in a storage area containing flammable and explosive materials, the preset abnormal behavior set may include "personnel illegally using open flames", "explosion-proof equipment failure", "material leakage", etc.; while in the office area, behaviors such as "personnel gathering and blocking the passage" and "fire-fighting facilities being blocked" may be concerned.

[0044] S200: Based on the multiple preset abnormal behavior sets, define the risk accompanying features and shielding features in the video images, and construct multiple primary video frame screening tools with the definition results.

[0045] Specifically, the risk - associated feature refers to an image feature that is closely related to a preset abnormal behavior and can indicate potential safety risks. For example, when it comes to the behavior of "person not wearing a safety helmet", features such as the shape and color of the safety helmet are risk - associated features. The shielding feature, on the other hand, refers to an image feature that is irrelevant to the abnormal behavior or should not trigger an alarm, such as fixed objects in the background, signs in a specific area, etc.

[0046] The primary video frame screening tool is a preliminary screening mechanism constructed based on the defined risk - associated features and shielding features, used to quickly filter a large amount of video data and extract video frames that may contain abnormal behaviors.

[0047] By defining the above - mentioned risk - associated features and shielding features, it is possible to efficiently exclude images that are ineffective for recognition, thereby helping to improve the overall response efficiency, while more accurately identifying abnormal behaviors in the video and reducing false alarms and missed alarms.

[0048] In some embodiments, based on the multiple preset abnormal behavior sets, the definition of risk - associated features and shielding features in video images is performed, and multiple primary video frame screening tools are constructed based on the definition results, including: For each abnormal behavior in the multiple preset abnormal behavior sets, obtain the warning shielding features defined by the user to generate the shielding feature definition result; for each abnormal behavior in the multiple preset abnormal behavior sets, collect the dynamic difference features of the video frames in the warning - triggered scenario and the non - triggered scenario to generate the risk - associated feature definition result; construct a dynamic detection model based on the risk - associated feature definition result and optimize it with the shielding feature definition result to generate the multiple primary video frame screening tools.

[0049] Specifically, for each preset abnormal behavior, first, obtain the warning shielding features defined by the user to generate the shielding feature definition result. Among them, the shielding feature is an image feature defined by the user according to the actual scenario, which has dynamic changes but should not trigger an alarm, and is used to suppress false alarms, such as the blowing of leaves by the wind, changes in light, regular operation of equipment, and infrared interference at night.

[0050] Then, collect the video segments and normal - state segments when the abnormal behavior occurs, conduct a comparative analysis, such as identifying the differences between the time periods with people in the video frame and the time periods without people (e.g., by using spatio - temporal pose estimation and optical flow analysis), and represent the difference information in vector form to generate the risk - associated feature definition result.

[0051] Furthermore, taking the risk - associated feature as the training target, construct a lightweight model (such as MobileNet, LSTM + optical flow), and use the shielding feature as a negative sample or a feature suppression factor to optimize the false - alarm rate of the model, generating multiple primary video frame screening tools.

[0052] Through the above steps, the preset abnormal behavior set can be transformed into a specific and operable detection model, providing a technical basis for subsequent video frame screening. Among them, the primary screening tool can run on the front-end or edge devices, quickly eliminating low-risk pictures from a large number of video frames, only retaining suspected abnormal frames, and reducing the recognition pressure on the back-end. Optimizing the model through user-defined shielding features can effectively exclude common interference sources, reduce false alarms, and improve the credibility of the system.

[0053] S300: Perform preliminary risk frame screening on the video acquisition results of the multiple video monitoring areas through the multiple primary video frame screening tools, generate multiple screened frame sets, and any screened frame carries a detection behavior identifier.

[0054] Specifically, the preliminary risk frame screening refers to using the primary video frame screening tool constructed above to conduct the first-round screening of a large amount of video data transmitted from each video monitoring area, and then classifying and summarizing the video frames determined by the screening tool as potentially having abnormal behaviors to form several sets, each set corresponding to a specific area, a specific type of abnormal behavior, etc. Among them, the video acquisition result is the video stream or the stored video file of each monitoring area.

[0055] Specifically, any screened frame carrying a detection behavior identifier means that each screened video frame will be marked with a label, which is used to label the identification information of the abnormal behavior type associated with the screened frame, such as "running", "falling", "illegal intrusion", etc., so that it can be quickly identified and classified during subsequent processing. Optionally, the detection behavior identifier is a category label, a behavior code, or a confidence score.

[0056] In this way, potential risk points can be quickly located from a large amount of video data, and different types of abnormal behaviors can be classified and summarized, providing an accurate and structured data basis for subsequent further analysis and processing, and improving the intelligent level and processing efficiency of the entire security assessment process.

[0057] S400: Analyze whether the data processing volume of the multiple screened frame sets is greater than a preset concurrent processing threshold.

[0058] Furthermore, based on the threshold method, determine whether the data volume of the multiple screened frame sets obtained in the previous step is too large (which can be measured by indicators such as the number of frames, data size, and time span). If the data volume is greater than the preset concurrent processing threshold, it can be considered that the current data volume to be processed is too large. Among them, the preset concurrent processing threshold is an upper limit value determined comprehensively according to various factors such as the system's hardware configuration, software processing capacity, and business requirements, and is used to limit the video frame data volume for in-depth analysis simultaneously to ensure that the system can operate in a stable and efficient state.

[0059] In the specific implementation process, first, relevant data of all screened frame sets are collected, such as the number of video frames in each set, the resolution and data size of each frame, etc. Then, the total data processing volume is obtained according to the preset calculation rules (such as simple addition of the number of frames, summation by data volume, or comprehensive evaluation combining time and space complexity, etc.). Next, this data processing volume is compared with the preset concurrent processing threshold. Thus, it is ensured that the processing flow of the entire system will not freeze or crash due to data overload, and the continuity and stability of the safety assessment work are ensured.

[0060] In some embodiments, the method for determining the preset concurrent processing threshold includes: Determine the deployment environment information of multiple primary video frame screening tools; perform twin simulation with the deployment environment information to determine the maximum stable concurrency number; generate the preset concurrent processing threshold with the maximum stable concurrency number.

[0061] In some implementation manners, the maximum stable concurrency number includes the frame rate that simultaneously satisfies the preset bottleneck usage rate of the CPU / GPU and the preset unstable condition.

[0062] Specifically, the preset concurrent processing threshold for scheduling control is determined through the twin simulation method to ensure the system stability and real-time performance when multiple primary video frame screening tools run concurrently. First, collect the hardware resource information of the current deployment platform, including but not limited to the number of CPU cores, GPU type and video memory capacity, memory size, I / O bandwidth, etc.; then, based on the above environment information, simulate the concurrent running scenarios of different numbers of screening tools in the digital twin environment, and evaluate the running stability of the system under different concurrency numbers; and according to the preset discrimination conditions, determine the maximum stable concurrency number, and use the maximum stable concurrency number as the concurrent processing threshold in the current deployment environment to provide a reference for the scheduling control of the system during operation.

[0063] Specifically, determining the maximum stable concurrency number is to determine the maximum number of concurrent running instances that can be supported on the premise of meeting the following two conditions: the system resource usage rate does not exceed the preset bottleneck threshold (such as CPU usage rate < 85%, GPU usage rate < 90%); the frame processing frame rate does not drop below the preset unstable condition (such as frame rate > 15fps).

[0064] Through the above mechanism, the concurrent threshold can be automatically set according to the deployment platform, avoiding overloading operation, improving the system stability and robustness, and at the same time facilitating the deployment and operation of this method on different platforms (such as edge devices, cloud platforms), that is, automatically adapting the performance parameters.

[0065] S500: If so, based on the detection behavior identifier, perform real-time risk analysis on each area, and call the multiple primary video frame screening tools according to the real-time risk priority to perform real-time security analysis and delayed security analysis on multiple screening frame sets, and generate multiple area security analysis results.

[0066] Specifically, if the data processing volume of multiple screening frame sets is greater than the preset concurrent processing threshold, it is necessary to perform step-by-step processing on multiple screening frame sets, that is, perform sequential security analysis (real-time security analysis and delayed security analysis) in the order of the risk priority of different screening frame sets. Among them, real-time security analysis is used to process the screening frames of high-risk or high-priority areas; delayed security analysis is used to process the screening frames of low-priority areas.

[0067] Through the above process, on the one hand, it ensures that high-risk areas can be promptly concerned and processed, effectively preventing accidents and protecting the safety of personnel's lives and property. On the other hand, it helps to reasonably utilize system resources, avoid resource waste and processing delays caused by indiscriminate real-time analysis of all areas, and improve the efficiency and accuracy of overall security assessment.

[0068] In some embodiments, based on the detection behavior identifier, perform real-time risk analysis on each area, and call the multiple primary video frame screening tools according to the real-time risk priority to perform real-time security analysis and delayed security analysis on multiple screening frame sets, and generate multiple area security analysis results, including: Perform the trigger risk evaluation of unsafe behaviors in each area based on the detection behavior identifier, and construct the real-time risk priority; combine the preset concurrent processing threshold and the real-time risk priority, match the real-time security analysis mode and the delayed security analysis mode for the multiple screening frame sets, and generate the real-time mode matching result and the delayed mode matching result; according to the real-time mode matching result and the delayed mode matching result, call the multiple primary video frame screening tools to perform security analysis on multiple screening frame sets, and generate the multiple area security analysis results.

[0069] Specifically, the real-time risk priority is the priority comprehensively evaluated according to factors such as the risk level of the detection behavior, the importance of the occurrence area, and the historical behavior frequency, and is used to guide the scheduling order of analysis resources.

[0070] Specifically, first, risk assessment is performed on each area according to a preset risk assessment model (constructed based on factors such as the type of abnormal behavior, the weight of the occurrence area, and the potential impact range). Exemplarily, this risk assessment model can be a rule-based regularization model. Exemplarily, the risk of "fighting" is higher than that of "wandering", such as the chemical plant area being higher than the office area. Then, the concurrent processing threshold of the current deployment environment is obtained (such as supporting a maximum of 5 real-time tasks), and the frame set is filtered according to the priority ranking; the high-priority tasks (the first 5 in the ranking result) are assigned to the real-time analysis mode, and the low-priority tasks (tasks 6 and later) are assigned to the delayed analysis mode, forming two task dispatch lists: the real-time mode matching result and the delayed mode matching result.

[0071] Furthermore, according to the matching results, multiple primary video frame screening tools are scheduled to perform real-time analysis on high-priority areas and delayed analysis on low-priority areas, and finally, all the analysis results are summarized to generate the security analysis results of multiple areas. For example, in a chemical industrial park, if the detection behavior identifier of a certain area shows "flammable liquid leakage", the real-time risk priority of this area will be higher than that of another area with only "person not wearing a safety helmet". The system will give priority to calling the screening tool to perform real-time analysis on the "flammable liquid leakage" area and perform delayed analysis on the "person not wearing a safety helmet" area.

[0072] Optionally, using the principle of the integrated learning method, with multiple primary video frame screening tools as weak learners, a corresponding integrated security analysis model is constructed to improve the security analysis performance and enhance the accuracy and robustness of risk identification.

[0073] Through the above process, risk-driven analysis resource scheduling can be achieved, ensuring that high-risk areas receive faster responses. At the same time, by queuing low-priority tasks in the delayed analysis mode, it is avoided that the system crashes due to task overload, improving the overall stability.

[0074] In some implementation manners, based on the detection behavior identifier, a trigger risk assessment of the unsafe behaviors in each area is performed to construct a real-time risk priority, including: Performing trigger feature fitting of the unsafe behaviors in each area based on the detection behavior identifier to construct multiple trigger features; performing a trigger risk assessment with the multiple trigger features, including the risk of the behavior itself and the risk of linkage and diffusion, to generate multiple risk evaluation values; arranging the multiple risk evaluation values in descending order to generate the real-time risk priority.

[0075] Specifically, first, for each detection behavior identifier, feature fitting is performed by combining the operating status of its affiliated area and the environmental background information to generate multiple trigger features including but not limited to the following information: behavior attributes (such as intrusion, occlusion, fall, crowding, etc.); trigger moment and periodic correlation; personnel density and activity frequency; geographical location of the affiliated area and connectivity of the surrounding areas; associated paths corresponding to historical risk events. Then, a dual-channel risk assessment is performed on the above multiple trigger features. Among them, the risk channel of the behavior itself scores dimensions such as behavior intensity, historical hazard index, and violation level; the linked diffusion risk channel evaluates the possibility of the behavior triggering secondary risks in adjacent areas under the current spatial structure (for example: a crowd stampede in one place may induce chaos in chain areas), and fuses the risk values of the two channels to obtain multiple risk evaluation values.

[0076] Further, the multiple risk evaluation values are sorted in descending order to form a risk priority queue that is updated in real time. Through the above steps, dynamic perception and risk ranking of unsafe behaviors in each area are achieved, and the risk propagation model and weight parameters can be dynamically adjusted according to different park structures and regional functions, thereby improving the environmental adaptability of this method.

[0077] In some embodiments, after generating multiple regional safety analysis results, it further includes: Reading a preset evaluation period; performing statistics on the safety analysis results of multiple video monitoring areas with the preset evaluation period to generate an unsafe behavior statistics table; performing safety assessments on the multiple video monitoring areas based on the unsafe behavior statistics table to generate multiple safety assessment results.

[0078] Specifically, first, obtain multiple safety assessment result data within the preset evaluation period as the basis for statistical analysis; then, within the set time period, summarize and organize the safety analysis results of all monitoring areas to form a detailed unsafe behavior statistics table, which records information such as the number of times and types of various unsafe behaviors that occurred in each area during this period; through the data in the statistics table, the safety status of each monitoring area can be comprehensively evaluated in combination with certain evaluation rules and standards, and finally multiple safety assessment results are generated to reflect the safety management status and existing problems of each area, providing a basis for subsequent improvement measures.

[0079] Exemplarily, a safety assessment is performed on the liquid ammonia and liquid chlorine areas of a certain park to monitor risk situations including single-person patrol phenomenon, the liquid chlorine warehouse door often being open, single-person operation, operation without a guardian, and incomplete wearing of inspection PPE. The corresponding unsafe behavior statistics table is as follows: Table 1 Unsafe Behavior Statistics Table for Liquid Ammonia and Liquid Chlorine Areas of a Certain Park Location Single-person inspection Warehouse door open / always open Single-person operation Operation without a guardian Incomplete wearing of PPE during inspection Liquid ammonia station 2 1 Liquid ammonia station 4 4 1 4 Liquid ammonia gasification area 3 4 Liquid ammonia loading and unloading area 1 2

[0080] In some implementations, after generating multiple security assessment results, it further includes: Read the continuous security assessment results under the preset assessment period, identify the high-frequency unsafe behaviors with the continuous repetition times of unsafe behaviors greater than the preset number and the corresponding high-frequency unsafe areas; retrieve the event occurrence videos corresponding to the high-frequency unsafe behaviors, and compare them with the monitoring videos corresponding to the low-frequency unsafe behaviors with the continuous repetition times less than the preset number, determine the unsafe factors, and send them to the management terminal for reminder.

[0081] Specifically, by analyzing the security assessment results of consecutive periods, it is possible to find out the unsafe behaviors that occur more than the set threshold in different periods and the specific areas where these behaviors occur. For example, if the preset number is 3 times, and the behavior of "operating the equipment in violation" appears 4 times in a certain area within three consecutive assessment periods, then this behavior is identified as a high-frequency unsafe behavior, and the corresponding area is the high-frequency unsafe area. Furthermore, retrieve the original video segments corresponding to the high-frequency unsafe behaviors, compare and analyze them with the videos of low-frequency unsafe behaviors, and determine the specific factors leading to the high-frequency unsafe behaviors through the detailed comparison of video frames (such as the operation methods of personnel, the status of equipment, environmental conditions, etc.), and then send this information to the management terminal so that the management personnel can timely understand the situation and take corresponding measures for improvement.

[0082] Through the above process, it is possible to achieve comprehensive, in-depth and dynamic monitoring and evaluation of the safety status of chemical industrial parks. On the one hand, the statistical and evaluation of periodic security analysis results can help management personnel timely understand the overall safety trend of the park and the safety management status of each area, discover potential safety hazards and management loopholes, so as to formulate improvement plans and training programs targeted, and improve the safety management level of the park. On the other hand, the identification and analysis of high-frequency unsafe behaviors can accurately locate typical problem areas and types, enabling management personnel to take more targeted measures to solve recurring safety problems, such as strengthening the supervision of specific areas, improving operation procedures or retraining relevant personnel, etc.

[0083] In summary, the chemical industrial park safety assessment method provided by the present invention combined with a video diagnosis tool has the following technical effects: By determining multiple preset abnormal behavior sets corresponding to multiple video monitoring areas within a target chemical industrial park; based on the multiple preset abnormal behavior sets, defining various risk accompanying features and shielding features, and constructing several primary video frame screening modules accordingly; using the multiple primary video frame screening modules to preliminarily screen the collected videos of each video monitoring area, extracting a set of risk frames containing detection behavior identifiers to form multiple preliminary screening frame sets; analyzing the data processing amounts required for the above multiple screening frame sets to determine whether they exceed a preset concurrent processing capacity threshold; if so, based on the detection behavior identifiers carried in the frames, performing real-time risk assessment for each monitoring area, sorting the call priorities of the primary video frame screening modules according to the assessment results, and respectively performing real-time safety analysis and delayed safety analysis to output corresponding regional safety analysis results, thereby achieving the technical effects of improving the monitoring data processing efficiency, enhancing the safety recognition accuracy and responsiveness.

[0084] Embodiment 2, such as Figure 2 is a schematic structural diagram of a chemical industrial park safety assessment system combining a video diagnosis tool according to the present invention. For example, Figure 1 In the present invention, the flowchart of the chemical industrial park safety assessment method combining a video diagnosis tool can be implemented through a structure such as Figure 2 shown.

[0085] Based on the same concept as the chemical industrial park safety assessment method combining a video diagnosis tool in the above embodiment, the chemical industrial park safety assessment system combining a video diagnosis tool provided by the present invention further includes: A behavior preset module 11, configured to determine multiple preset abnormal behavior sets of multiple video monitoring areas within a target chemical industrial park.

[0086] A screening tool construction module 12, configured to define risk accompanying features and shielding features in video images based on the multiple preset abnormal behavior sets, and construct multiple primary video frame screening tools based on the definition results.

[0087] A preliminary risk frame screening module 13, configured to perform preliminary risk frame screening on the video acquisition results of the multiple video monitoring areas through the multiple primary video frame screening tools, generate multiple screening frame sets, and any screening frame carries a detection behavior identifier.

[0088] A processing amount analysis module 14, configured to analyze whether the data processing amounts of the multiple screening frame sets are greater than a preset concurrent processing threshold.

[0089] A safety analysis module 15, configured to, if so, perform real-time risk analysis for each area based on the detection behavior identifier, and call the multiple primary video frame screening tools according to the real-time safety priority to perform real-time safety analysis and delayed safety analysis on the multiple screening frame sets, and generate multiple regional safety analysis results.

[0090] Further, in the behavior preset module 11, any one of the multiple preset abnormal behavior sets includes unsafe behaviors of people and unsafe states of objects.

[0091] In some embodiments, the screening tool construction module 12 includes: An early warning shielding feature acquisition and definition unit, configured to acquire, for each abnormal behavior in the multiple preset abnormal behavior sets, the early warning shielding features defined by the user, and generate a shielding feature definition result.

[0092] A risk accompanying feature collection and definition unit, configured to collect, for each abnormal behavior in the multiple preset abnormal behavior sets, the dynamic difference features of the pictures in the early warning trigger scenario and the non-trigger scenario, and generate a risk accompanying feature definition result.

[0093] A primary video frame screening tool generation unit, configured to construct a dynamic detection model based on the risk accompanying feature definition result, and optimize it with the shielding feature definition result to generate the multiple primary video frame screening tools.

[0094] In some embodiments, the safety analysis module 15 includes: A risk real-time priority construction unit, configured to perform a triggering risk evaluation of the unsafe behaviors in each area based on the detection behavior identifier, and construct a risk real-time priority.

[0095] A safety analysis mode matching unit, configured to combine the preset concurrency processing threshold and the risk real-time priority to match the multiple screening frame sets with the real-time safety analysis mode and the delayed safety mode, and generate a real-time mode matching result and a delayed mode matching result.

[0096] A regional safety analysis result generation unit, configured to, according to the real-time mode matching result and the delayed mode matching result, call the multiple primary video frame screening tools to perform a safety analysis on the multiple screening frame sets, and generate the multiple regional safety analysis results.

[0097] In some implementation manners, the risk real-time priority construction unit in the safety analysis module 15 includes:

[0098] A trigger feature fitting subunit, configured to perform a trigger feature fitting of the unsafe behaviors in each area based on the detection behavior identifier, and construct multiple trigger features.

[0099] A trigger risk evaluation subunit, configured to perform a trigger risk evaluation with the multiple trigger features, including the risk of the behavior itself and the risk of linkage and diffusion, and generate multiple risk evaluation values.

[0100] A risk real-time priority generation subunit, configured to sort the multiple risk evaluation values in descending order to generate the risk real-time priority.

[0101] In some embodiments, the throughput analysis module 14 includes: A deployment environment information determination unit, configured to determine the deployment environment information of multiple primary video frame screening tools.

[0102] A twin simulation and maximum stable concurrency determination unit, configured to perform twin simulation based on the deployment environment information to determine the maximum stable concurrency.

[0103] A preset concurrency processing threshold generation unit, configured to generate the preset concurrency processing threshold based on the maximum stable concurrency.

[0104] Furthermore, the maximum stable concurrency includes the frame rate that simultaneously satisfies the preset bottleneck usage rate of the CPU / GPU and the preset unstable conditions.

[0105] In some embodiments, the chemical industrial park safety assessment system combined with the video diagnosis tool further includes: A preset evaluation period reading unit, configured to read the preset evaluation period.

[0106] A safety analysis result statistics unit, configured to perform statistics on the safety analysis results of multiple video monitoring areas based on the preset evaluation period to generate an unsafe behavior statistics table.

[0107] A safety assessment result generation unit, configured to perform safety assessment on the multiple video monitoring areas based on the unsafe behavior statistics table to generate multiple safety assessment results.

[0108] In some implementation manners, the safety assessment result generation unit includes: A high-frequency unsafe behavior recognition subunit, configured to read the consecutive safety assessment results under the preset evaluation period, and recognize high-frequency unsafe behaviors whose consecutive repetition times of unsafe behaviors are greater than the preset number and the corresponding high-frequency unsafe areas.

[0109] An unsafe factor determination and reminder subunit, configured to retrieve the event occurrence videos corresponding to the high-frequency unsafe behaviors, and compare them with the monitoring videos corresponding to the low-frequency unsafe behaviors whose consecutive repetition times are less than the preset number, determine the unsafe factors, and send them to the management end for reminder.

[0110] It should be understood that the embodiments mentioned in this specification focus on their differences from other embodiments. The specific embodiments in the foregoing Embodiment 1 are equally applicable to the chemical industrial park safety assessment system combined with the video diagnosis tool described in Embodiment 2. For the sake of brevity of the specification, no further elaboration is made here.

[0111] It should be understood that the disclosed embodiments of the present invention and the above descriptions enable those skilled in the art to implement the present invention using the present invention. At the same time, the present invention is not limited to the above-mentioned part of the embodiments. It should be understood that ordinary skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A safety assessment method for chemical industrial parks combined with video diagnostic tools, characterized in that Including: Determine multiple preset abnormal behavior sets for multiple video monitoring areas within the target chemical industrial park; Based on the multiple preset abnormal behavior sets, define risk accompanying features and shielding features in the video image, and construct multiple primary video frame screening tools with the definition results; Perform preliminary risk frame screening on the video acquisition results of the multiple video monitoring areas through the multiple primary video frame screening tools, generate multiple screened frame sets, and any screened frame carries a detection behavior identifier; Analyze whether the data processing volume of the multiple screened frame sets is greater than a preset concurrent processing threshold; If so, perform real-time risk analysis for each area based on the detection behavior identifier, and call the multiple primary video frame screening tools according to the real-time risk priority to perform real-time security analysis and delayed security analysis on the multiple screened frame sets, and generate multiple area security analysis results.

2. The chemical industrial park safety assessment method combined with a video diagnosis tool according to claim 1, wherein After generating multiple area security analysis results, it further includes: Read the preset evaluation period; Perform statistics on the security analysis results of the multiple video monitoring areas with the preset evaluation period, and generate an unsafe behavior statistics table; Perform security evaluation on the multiple video monitoring areas based on the unsafe behavior statistics table, and generate multiple security evaluation results.

3. The chemical industrial park safety assessment method combined with a video diagnosis tool according to claim 1, wherein Any one of the multiple preset abnormal behavior sets includes unsafe behaviors of people and unsafe states of objects.

4. The chemical industrial park safety assessment method combined with a video diagnosis tool according to claim 1, wherein Based on the multiple preset abnormal behavior sets, define risk accompanying features and shielding features in the video image, and construct multiple primary video frame screening tools, including: For each abnormal behavior in the multiple preset abnormal behavior sets, obtain the warning shielding features defined by the user, and generate shielding feature definition results; For each abnormal behavior in the multiple preset abnormal behavior sets, collect the dynamic difference features of the screen in the warning trigger scenario and the non-trigger scenario, and generate risk accompanying feature definition results; Construct a dynamic detection model with the risk accompanying feature definition results, and optimize it with the shielding feature definition results to generate the multiple primary video frame screening tools.

5. The chemical industrial park safety assessment method combined with a video diagnosis tool according to claim 1, characterized in that Perform real-time risk analysis for each area based on the detection behavior identifier, and call the multiple primary video frame screening tools according to the real-time risk priority to perform real-time security analysis and delayed security analysis on the multiple screened frame sets, and generate multiple area security analysis results, including: Perform trigger risk evaluation on the unsafe behaviors in each area based on the detection behavior identifier, and construct a real-time risk priority; Combine the preset concurrent processing threshold and the real-time risk priority, match the real-time security analysis mode and the delayed security mode for the multiple screened frame sets, and generate a real-time mode matching result and a delayed mode matching result; According to the real-time mode matching result and the delayed mode matching result, call the multiple primary video frame screening tools to perform security analysis on the multiple screened frame sets, and generate the multiple area security analysis results.

6. The chemical industrial park safety assessment method combined with a video diagnosis tool according to claim 5, characterized in that, Perform trigger risk evaluation on the unsafe behaviors in each area based on the detection behavior identifier, and construct a real-time risk priority, including: Perform trigger feature fitting on the unsafe behaviors in each area based on the detection behavior identifier, and construct multiple trigger features; Trigger risk assessment is performed based on the multiple trigger features, including the risk of the behavior itself and the risk of linkage spread, to generate multiple risk evaluation values; Arrange the multiple risk evaluation values in descending order to generate the real-time risk priority.

7. The chemical industrial park safety assessment method combined with a video diagnosis tool according to claim 1, wherein The method for determining the preset concurrency processing threshold includes: Determine the deployment environment information of multiple primary video frame screening tools; Perform twin simulation based on the deployment environment information to determine the maximum stable concurrency; Generate the preset concurrency processing threshold based on the maximum stable concurrency.

8. The chemical industrial park safety assessment method combined with a video diagnosis tool according to claim 7, characterized in that, The maximum stable concurrency includes the frame rate that simultaneously meets the preset bottleneck usage rate of the CPU / GPU and the preset unstable conditions.

9. The chemical industrial park safety assessment method combined with a video diagnosis tool according to claim 2, characterized in that, After generating multiple security assessment results, it further includes: Read the continuous security assessment results under the preset assessment period, and identify the high-frequency unsafe behaviors and the corresponding high-frequency unsafe areas where the continuous repetition times of the unsafe behaviors are greater than the preset times; Retrieve the event occurrence videos corresponding to the high-frequency unsafe behaviors, compare them with the monitoring videos corresponding to the low-frequency unsafe behaviors with continuous repetition times less than the preset times, determine the unsafe factors, and send them to the management end for reminder.

10. A chemical industrial park safety assessment system combined with a video diagnosis tool, characterized in that, A chemical industrial park security assessment method for implementing the combined video diagnosis tool according to any one of claims 1 to 9 includes: A behavior preset module for determining multiple preset abnormal behavior sets in multiple video monitoring areas within the target chemical industrial park; A screening tool construction module for defining the risk accompanying features and shielding features in the video image based on the multiple preset abnormal behavior sets, and constructing multiple primary video frame screening tools based on the definition results; A preliminary risk frame screening module for performing preliminary risk frame screening on the video acquisition results of the multiple video monitoring areas through the multiple primary video frame screening tools to generate multiple screening frame sets, and any screening frame carries a detection behavior identifier; A processing volume analysis module for analyzing whether the data processing volume of the multiple screening frame sets is greater than the preset concurrency processing threshold; A security analysis module for, if so, performing real-time risk analysis on each area based on the detection behavior identifier, and invoking the multiple primary video frame screening tools according to the real-time risk priority to perform real-time security analysis and delayed security analysis on the multiple screening frame sets, and generating multiple area security analysis results.

Images