Smelting Hidden Danger Warning Method and System Based on Video Data Analysis

Abstract

The present invention discloses a method and system for warning of smelting hazards based on video data analysis, relating to the technical field of early warning and control in smelting factories. Specifically, it discloses dividing monitoring blocks for the video images captured by cameras and determining monitoring strategies according to monitoring objects; analyzing the video images of the monitoring blocks to identify equipment states or human behaviors, and evaluating abnormal situations using the monitoring strategies; constructing a floor plan of the smelting site, marking the monitoring blocks, and performing correlation analysis based on the characteristics of equipment interlock effects to monitor abnormal situations in real time, marking the abnormal blocks, and analyzing their interlock effects to determine the degree of hazard warning; determining the corresponding hazard alarm level according to the interval to which the degree of hazard warning belongs. The above technical solution of the present invention effectively improves the ability to identify safety hazards and the emergency response speed in the smelting site through refined monitoring, real-time analysis, interlock effect analysis, and determination of alarm levels.

Description

Technical Field

[0001] The present invention relates to the technical field of warning and control in smelting factories, and in particular to a method and system for warning of smelting hazards based on video data analysis. Background Art

[0002] In the production process of the smelting industry, safety management is a crucial link. However, the traditional smelting site monitoring system has many limitations and cannot meet the safety management requirements of modern smelting sites.

[0003] Traditional monitoring systems often adopt a global or fixed-area monitoring method, lacking refined monitoring of specific equipment or personnel operation areas. This results in inaccurate monitoring ranges and the inability to effectively capture the activities of key equipment and personnel. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system that can perform refined monitoring on the equipment and staff in a smelting factory.

[0005] The present invention discloses a method for warning of smelting hazards based on video data analysis, including:

[0006] Dividing the monitoring blocks for the video images captured by each camera, and determining the monitoring strategy for each monitoring block based on the objects to be monitored in the monitoring block;

[0007] Analyzing the video images corresponding to each monitoring block to determine the status description of the equipment or the behavior description of the human body corresponding to the monitoring block, and analyzing the status description or behavior description using the monitoring strategy corresponding to each monitoring block to determine the abnormal situation assessment of the corresponding equipment or human body;

[0008] Constructing a floor plan of the smelting site, marking each monitoring block on the floor plan of the smelting site, associating different monitoring blocks based on the chain influence characteristics between equipment, monitoring the abnormal situation assessment of each monitoring block in real time, marking the monitoring blocks whose abnormal situation assessment is determined to be abnormal, and analyzing the chain influence relationship of the retrieved blocks with abnormal marks, and determining the warning level of the hidden dangers in the smelting site based on the analysis results;

[0009] Determining the hidden danger alarm level based on the hidden danger warning level interval to which the hidden danger warning level belongs.

[0010] In some embodiments disclosed by the present invention, the method for dividing the monitoring blocks in the video images includes:

[0011] Determining the key equipment areas and personnel operation areas in the video images, and dividing the video images based on the key equipment areas and personnel operation areas.

[0012] In some embodiments disclosed by the present invention, the method for determining the monitoring strategy of a monitoring block based on the object to be monitored in the monitoring block includes:

[0013] Construct a number of abnormal state types for the abnormal states in the critical equipment area, and for each abnormal state type, construct a number of abnormal state levels, and each abnormal state level corresponds to a number of abnormal state judgment factor parameter intervals;

[0014] Define a number of personnel sub-operation areas in the personnel operation area, and for the abnormal human behaviors in each personnel sub-operation area, construct a number of abnormal behavior types, and for each abnormal behavior type, construct a number of abnormal behavior levels, and each abnormal behavior level corresponds to a number of abnormal behavior judgment factor parameter intervals.

[0015] In some embodiments disclosed by the present invention, the method for analyzing the video image corresponding to each monitoring block to determine the state description of the equipment corresponding to the monitoring block includes:

[0016] Define the characteristic edges or mark the characteristic signs of the key structures of the equipment in the monitoring block;

[0017] Based on the normal operation state of the equipment in the monitoring block, determine the movement ranges of different characteristic edges or characteristic signs, and conduct several levels of sub-movement range delineation for the movement ranges, and configure a first abnormal movement parameter for each sub-movement range;

[0018] Based on the normal operation state of the equipment in the monitoring block, determine the movement speed intervals of different characteristic edges or characteristic signs, and conduct several levels of sub-movement speed interval delineation for the movement speed intervals, and configure a second abnormal movement parameter for each sub-movement speed interval;

[0019] Recognize the first abnormal movement parameter and the second abnormal movement parameter as the state description of the equipment in the monitoring block.

[0020] In some embodiments disclosed by the present invention, the method for determining the behavior description of the human body in the monitoring block includes:

[0021] Use human body recognition technology to perform real-time recognition of the human body in the monitoring block, and construct a real-time human body limb expression model for the recognized human body;

[0022] Determine different personnel working areas in the monitoring block, and determine a number of normal human postures for each personnel working area. Based on the normal human postures, construct a normal human body limb expression model, and record the combination of a number of normal human body limb expression models as the normal human body limb expression model group;

[0023] The real-time human body limb expression model is compared and analyzed using the normal human body limb expression model group. If no matching normal human body limb expression model is found in the normal human body limb expression model group, the real-time human body limb expression model is marked as abnormal;

[0024] Count the number of abnormal behaviors of the real-time human body limb expression model marked as abnormal within a preset time period, and determine the behavior description of the human body in the monitored area based on the number of abnormal behaviors.

[0025] In some embodiments disclosed by the present invention, the method for analyzing the status description or behavior description using the monitoring strategy corresponding to each monitoring area to determine the abnormal situation evaluation of the corresponding device or human body includes:

[0026] Perform real-time motion analysis on the feature edges or feature marks in the monitoring area to determine the first abnormal motion parameter and the second abnormal motion parameter of the device, and determine the comprehensive abnormal motion parameter of the device corresponding to the monitoring area based on the first abnormal motion parameter and the second abnormal motion parameter;

[0027] Among them, the expression for calculating the comprehensive abnormal motion parameter is:

[0028] ;

[0029] Among them, is the comprehensive abnormal motion parameter, is the influence weight of the first abnormal motion parameter, is the first abnormal motion parameter, is the influence weight of the second abnormal motion parameter, is the second abnormal motion parameter, is the continuous abnormal state duration of the device, is the influence adjustment coefficient of the continuous abnormal state duration, and c is the influence adjustment constant of the continuous abnormal state duration;

[0030] For the number of abnormal behaviors of the real-time human body limb expression model, several intervals of the number of abnormal behaviors are constructed, and a human body abnormal parameter is set for each interval of the number of abnormal behaviors. By judging the interval of the number of abnormal behaviors to which the number of abnormal behaviors belongs, the human body abnormal parameter of the real-time human body limb expression model is determined.

[0031] In some embodiments disclosed by the present invention, the method for analyzing the chain influence relationship of the retrieved area with abnormal marks and determining the hidden danger warning level of the smelting site based on the analysis result includes:

[0032] Determine the comprehensive abnormal motion parameters of the devices corresponding to different monitoring blocks, or the human body abnormal parameters of the human body. If the comprehensive abnormal motion parameters or the human body abnormal parameters are greater than or equal to the preset value, mark the monitoring block as abnormal;

[0033] Determine the total number of abnormal blocks in the monitoring blocks marked as abnormal, and determine the number of chain abnormal blocks in the monitoring blocks marked as abnormal that have a chain relationship, and calculate the chain abnormal ratio of the number of chain abnormal blocks to the total number of abnormal blocks;

[0034] Based on the comprehensive abnormal motion parameters of different monitoring blocks, or the human body abnormal parameters of the human body, and the chain abnormal ratio, determine the hidden danger warning level of the smelting site.

[0035] In some embodiments disclosed in the present invention, the expression for determining the hidden danger warning level of the smelting site is:

[0036] ;

[0037] Wherein, is the hidden danger warning level, is the number of chain abnormal blocks, is the total number of abnormal blocks, is the influence adjustment coefficient of the chain abnormal ratio, is the influence adjustment constant of the chain abnormal ratio, is the comprehensive abnormal motion parameter of the i-th monitoring block marked as abnormal, is the human body abnormal parameter of the i-th monitoring block marked as abnormal, and n is the total number of monitoring blocks marked as abnormal.

[0038] In some embodiments disclosed in the present invention, there is also disclosed a smelting hidden danger warning system based on video data analysis, including:

[0039] The first module is used to divide the video images captured by each camera into monitoring blocks, and determine the monitoring strategy of the monitoring block based on the object to be monitored in the monitoring block;

[0040] The second module is used to analyze the video images corresponding to each monitoring block, determine the state description of the device corresponding to the monitoring block or the behavior description of the human body, and analyze the state description or behavior description by using the monitoring strategy corresponding to each monitoring block to determine the abnormal situation evaluation of the corresponding device or human body;

[0041] The third module is used to construct a smelting site floor plan, mark each monitoring block on the smelting site floor plan, based on the interlock impact characteristics between equipment rooms, associate different monitoring blocks for interlock impact, monitor the abnormal situation assessment of each monitoring block in real time, mark the monitoring blocks whose abnormal situation assessment is identified as abnormal, analyze the interlock impact relationship of the retrieved blocks with abnormal marks, and based on the analysis results, determine the hidden danger warning level of the smelting site;

[0042] The fourth module is used to determine the hidden danger alarm level based on the hidden danger warning level interval to which the hidden danger warning level belongs.

[0043] The present invention discloses a method and system for smelting hidden danger warning based on video data analysis, which relates to the technical field of early warning and control of smelting factories. Specifically, it discloses dividing monitoring blocks for the video images captured by cameras and determining monitoring strategies according to monitoring objects; analyzing the video images of the monitoring blocks, identifying equipment states or human behaviors, and evaluating abnormal situations using the monitoring strategies; constructing a smelting site floor plan, marking the monitoring blocks, and conducting correlation analysis based on the equipment interlock impact characteristics, monitoring abnormal situations in real time, marking the abnormal blocks, and analyzing their interlock impacts to determine the hidden danger warning level; determining the corresponding hidden danger alarm level according to the interval to which the hidden danger warning level belongs; the above technical solutions of the present invention effectively improve the ability to identify safety hidden dangers and the emergency response speed of the smelting site through refined monitoring, real-time analysis, interlock impact analysis, and alarm level determination.

[0044] The technical solutions of the present invention will be further described in detail below through the drawings and embodiments. Description of the Drawings

[0045] Figure 1 It is a method step diagram of the smelting hidden danger warning method based on video analysis disclosed in the embodiment of the present invention. Detailed Embodiments

[0046] The technical solutions of the present invention will be further described below through the drawings and embodiments.

[0047] The technical solutions of the present invention will be clearly and completely described below in combination with the drawings and specific embodiments. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention and cannot be construed as limiting the protection scope of the present invention. Those skilled in the art can make some non-essential improvements and adjustments based on the content of the present invention described below. In the present invention, unless otherwise clearly defined and limited, the technical terms used in the present invention should have the ordinary meaning understood by those skilled in the art of the present invention.

[0048] Embodiment:

[0049] The present invention discloses a smelting hazard warning method based on video data analysis. Refer to Figure 1 , including:

[0050] Step S100, divide the monitoring blocks for the video images captured by each camera, and determine the monitoring strategy for the monitoring blocks based on the objects to be monitored in the monitoring blocks.

[0051] In the smelting site, due to the large number of devices and frequent personnel flow, a single camera often has difficulty covering all key areas. Therefore, first, it is necessary to reasonably divide the shooting range of each camera to form several monitoring blocks. These monitoring blocks can be determined based on factors such as device location, personnel activity area, or potential safety hazard points. Then, according to the objects to be monitored in each monitoring block (such as device operating status, personnel behavior, etc.), corresponding monitoring strategies are formulated. The monitoring strategies may include monitoring frequency, analysis algorithms, abnormal determination criteria, etc., to ensure that abnormal situations within the monitoring blocks can be accurately and efficiently captured.

[0052] In some embodiments disclosed by the present invention, the method for dividing the monitoring blocks in the video images includes:

[0053] Step S101, determine the key device areas and personnel operation areas in the video images, and divide the video images based on the key device areas and personnel operation areas.

[0054] In a complex environment such as a smelting site, the video images contain a large amount of device and personnel information. To monitor more effectively, first, it is necessary to determine the key device areas and personnel operation areas in the video images. The key device areas usually refer to the areas where the devices that are crucial to the production process and may cause serious consequences once a failure occurs are located. The personnel operation areas refer to the areas where personnel often carry out operations, operations, or activities.

[0055] Based on these key device areas and personnel operation areas, the video images can be divided into blocks. The principle of division can be based on the type, location, function of the devices or the activity range, operation content, etc. of the personnel. Through block division, the video images can be segmented into several relatively independent areas with clear monitoring objectives, providing a basis for formulating subsequent monitoring strategies.

[0056] In some embodiments disclosed by the present invention, the method for determining the monitoring strategy for the monitoring blocks based on the objects to be monitored in the monitoring blocks includes:

[0057] Step S102, construct several abnormal state types for the abnormal states of the key device areas, and for each abnormal state type, construct several abnormal state levels, and each abnormal state level corresponds to several abnormal state judgment factor parameter intervals.

[0058] In the present invention, special attention is paid to the anomalies in the movement range and movement speed of the equipment as the types of abnormal states in the key equipment area. An anomaly in the movement range may refer to the equipment exceeding its normal working range or movement trajectory, while an anomaly in the movement speed may refer to the equipment moving too fast or too slow, not conforming to its normal working speed.

[0059] For these two types of abnormal states, several levels of abnormal states are further constructed. The levels of abnormal states can be divided according to the severity of the anomaly, the degree of impact on production, or the possible consequences. For example, a slight excess in the movement range may only be a small deviation of the equipment, while a severe excess in the movement range may lead to equipment damage or production accidents.

[0060] For each level of abnormal state, several intervals of abnormal state judgment factor parameters are corresponding. For the anomaly in the movement range, the judgment factor parameter may be the deviation value between the actual position of the equipment and its normal working range; for the anomaly in the movement speed, the judgment factor parameter may be the difference or ratio between the actual movement speed of the equipment and its normal working speed.

[0061] Step S103, several personnel sub-operation areas are delimited in the personnel operation area, and several types of abnormal human behaviors are constructed for each personnel sub-operation area, and for each type of abnormal behavior, several levels of abnormal behavior are constructed, and each level of abnormal behavior corresponds to several intervals of abnormal behavior judgment factor parameters.

[0062] After determining the types and levels of abnormal states, this information is incorporated into the monitoring strategy. For the key equipment area, the monitoring system will continuously monitor the movement range and movement speed of the equipment and compare them with the preset intervals of abnormal state judgment factor parameters.

[0063] Once the movement range or movement speed of the equipment exceeds the preset parameter interval, the monitoring system will immediately trigger an abnormal alarm and take corresponding countermeasures according to the level of abnormal state. For example, for a slight excess in the movement range, it may only issue a warning signal to alert the operator; while for a severe excess in the movement range or abnormal movement speed, it may be necessary to immediately stop the equipment operation to prevent the situation from deteriorating further.

[0064] Step S200, analyze the video image corresponding to each monitoring block, determine the state description of the equipment or the behavior description of the human body corresponding to the monitoring block, and use the monitoring strategy corresponding to each monitoring block to analyze the state description or behavior description to determine the corresponding abnormal situation assessment of the equipment or the human body.

[0065] After determining the monitoring blocks and monitoring strategies, the next step is to conduct real-time analysis of the video images corresponding to each monitoring block. This step primarily relies on computer vision technology and image processing algorithms. By analyzing the video images, it is possible to extract device status descriptions (such as whether the equipment is operating normally or whether there is any abnormal vibration) or human behavior descriptions (such as whether a person is operating in an illegal manner or is in a dangerous area) within the monitoring block. These status or behavior descriptions are then analyzed using the previously determined monitoring strategy to assess whether any anomalies exist. If the assessment results indicate an anomaly, further action is required.

[0066] In some embodiments disclosed herein, a method for analyzing a video image corresponding to each monitoring block to determine a status description of a device corresponding to the monitoring block includes:

[0067] Step S201 : Delineate the feature edges or mark the feature marks of the key structures of the equipment in the monitoring block.

[0068] Step S202: Based on the normal operation status of the equipment in the monitoring block, the motion ranges of different characteristic edges or characteristic marks are determined, and several levels of sub-motion ranges are delineated for the motion ranges, and a first abnormal motion parameter is configured for each sub-motion range.

[0069] When the device is in normal operation, the movement range of its characteristic edge or characteristic mark is relatively stable. Therefore, by observing and analyzing the movement trajectory of the device during normal operation, its movement range can be determined. In order to describe the movement state of the device more finely, we divide this movement range into several sub-motion ranges, each of which represents a specific area that the device may reach during normal operation. At the same time, a first abnormal movement parameter is configured for each sub-motion range to describe the abnormal movement of the device within the range. When the movement of the device exceeds a certain sub-motion range, the first abnormal movement parameter is triggered, indicating that the device may have an abnormal movement range, thereby triggering an alarm in the monitoring system in a timely manner.

[0070] Step S203, based on the normal operating status of the equipment in the monitoring block, the motion speed intervals of different characteristic edges or characteristic marks are determined, and several levels of sub-motion speed intervals are delineated for the motion speed intervals, and a second abnormal motion parameter is configured for each sub-motion speed interval.

[0071] When the device is in normal operation, the movement speed of its characteristic edges or feature marks is also relatively stable. By observing and analyzing the movement speed of the device during normal operation, its movement speed range can be determined. To more accurately describe the movement state of the device, this movement speed range is divided into several sub-movement speed ranges, and each sub-movement speed range represents a specific speed range that the device may reach during normal operation. At the same time, a second abnormal movement parameter is configured for each sub-movement speed range to describe the abnormal movement of the device within this speed range. When the movement speed of the device exceeds a certain sub-movement speed range, the second abnormal movement parameter is triggered, indicating that the device may have an abnormal movement speed, thereby promptly triggering an alarm in the monitoring system.

[0072] Step S204, identify the first abnormal movement parameter and the second abnormal movement parameter as the state description of the device in the monitoring block.

[0073] In the monitoring system, the state description of the device is an important basis for judging whether the device is operating normally. By combining the first abnormal movement parameter and the second abnormal movement parameter, a comprehensive state description of the device can be obtained. These two parameters respectively reflect the abnormal conditions of the device in terms of movement range and movement speed. When either abnormal movement parameter of the device is triggered, it indicates that the movement state of the device is abnormal, and the monitoring system can issue an alarm signal accordingly to remind the operator to take timely measures for handling. Therefore, identifying the first abnormal movement parameter and the second abnormal movement parameter as the state description of the device in the monitoring block is a key step in realizing the timely discovery and handling of the abnormal state of the device.

[0074] In some embodiments disclosed by the present invention, the method for determining the behavior description of a human body within a monitoring block includes:

[0075] Step S205, using human body recognition technology, perform real-time recognition on the human body in the monitoring block, and construct a real-time human body limb expression model for the recognized human body.

[0076] In the monitoring system, in order to accurately describe and analyze the behavior of a human body, it is first necessary to use human body recognition technology to perform real-time recognition on the human body in the monitoring block. Through image processing and machine learning algorithms, the system can automatically detect and track the human body in the video, and extract key feature points of the human body, such as joint positions, limb directions, etc. Based on these feature points, the system constructs a real-time human body limb expression model, which can reflect the posture and actions of the human body in real time and provide a basis for subsequent behavior analysis.

[0077] Step S206: Determine the working areas of different personnel in the monitoring block, determine several normal human postures for each personnel's working area, construct a normal human limb expression model based on the normal human postures, and denote the combination of several normal human limb expression models as the normal human limb expression model group.

[0078] Within the monitoring block, the working areas and responsibilities of different personnel may vary, so their normal human postures will also be different. To accurately describe the normal behaviors of the human body, the system first needs to determine the working areas of different personnel in the monitoring block and observe and record the normal human postures within each working area. Based on these normal postures, the system constructs a normal human limb expression model, which can reflect the postures and movements of the human body in a normal working state. At the same time, the system combines several normal human limb expression models to form a normal human limb expression model group for subsequent behavior comparison and analysis.

[0079] Step S207: Compare and analyze the real-time human limb expression model with the normal human limb expression model group. If no matching normal human limb expression model is found in the normal human limb expression model group, mark the real-time human limb expression model as abnormal.

[0080] After obtaining the real-time human limb expression model and the normal human limb expression model group, the system needs to compare and analyze them. By calculating the similarity between the real-time human limb expression model and each model in the normal human limb expression model group, the system can determine whether the real-time human posture is normal. If no model in the normal human limb expression model group matches the real-time human limb expression model, that is, the similarity is lower than a certain threshold, then the system considers the real-time human posture to be abnormal and marks it.

[0081] Step S208: Count the number of abnormal behaviors of the real-time human limb expression models marked as abnormal within a preset time period, and determine the behavior description of the human body in the monitoring block based on the number of abnormal behaviors.

[0082] To comprehensively describe the behaviors of the human body within the monitoring block, the system needs to count the number of abnormal behaviors of the real-time human limb expression models marked as abnormal within a preset time period. By recording and analyzing these numbers of abnormal behaviors, the system can understand the abnormal behavior situations of the human body within the monitoring block. Based on these numbers of abnormal behaviors, the system can determine the behavior description of the human body within the monitoring block, such as "frequent abnormal behaviors", "occasional abnormal behaviors", or "normal behaviors", etc. Such behavior descriptions help the monitoring system to promptly detect and handle the abnormal behaviors of the human body and ensure the safety and order of the monitoring area.

[0083] In some embodiments disclosed by the present invention, a method for analyzing a status description or a behavior description by using a monitoring strategy corresponding to each monitoring block to determine an abnormal condition evaluation of a corresponding device or a human body includes:

[0084] Step S209: Perform real-time motion analysis on the feature edges or feature marks in the monitoring block to determine a first abnormal motion parameter and a second abnormal motion parameter of the device, and based on the first abnormal motion parameter and the second abnormal motion parameter, determine a comprehensive abnormal motion parameter of the device corresponding to the monitoring block.

[0085] Among them, the expression for calculating the comprehensive abnormal motion parameter is:

[0086] .

[0087] Among them, is the comprehensive abnormal motion parameter, is the influence weight of the first abnormal motion parameter, is the first abnormal motion parameter, is the influence weight of the second abnormal motion parameter, is the second abnormal motion parameter, is the continuous abnormal state duration of the device, is the influence adjustment coefficient of the continuous abnormal state duration, and c is the influence adjustment constant of the continuous abnormal state duration.

[0088] Step S2010: Construct a number of abnormal behavior times intervals for the abnormal behavior times of the real-time human body limb expression model, and set human body abnormal parameters for each abnormal behavior times interval. By judging the abnormal behavior times interval to which the abnormal behavior times belong, determine the human body abnormal parameters of the real-time human body limb expression model.

[0089] Step S300: Construct a floor plan of the smelting site, mark each monitoring block on the floor plan of the smelting site, based on the chain influence characteristics between devices, perform chain influence association on different monitoring blocks, perform real-time monitoring on the abnormal condition evaluation of each monitoring block, mark the monitoring blocks whose abnormal condition evaluation is determined to be abnormal, perform chain influence relationship analysis on the retrieved blocks with abnormal marks, and based on the analysis results, determine the hidden danger warning level of the smelting site.

[0090] First, it is necessary to construct a floor plan of the smelting site and mark each monitoring block on the floor plan. This helps to intuitively understand the distribution and positional relationship of the monitoring blocks. Then, based on the chain effect characteristics between devices (for example, the failure of a certain device may cause the shutdown or damage of other devices), different monitoring blocks are associated with chain effects. Next, real-time monitoring is carried out on the abnormal situation assessment of each monitoring block. Once a monitoring block is evaluated as abnormal, it is immediately marked as abnormal and the chain effect relationship analysis is triggered. By analyzing the chain effect relationship between the abnormal monitoring block and other monitoring blocks, the hidden danger warning level of the entire smelting site is evaluated. This step helps to detect potential safety hazards in advance and avoid accidents.

[0091] In some embodiments disclosed in the present invention, the method for analyzing the chain effect relationship of the retrieved block with an abnormal mark and determining the hidden danger warning level of the smelting site based on the analysis result includes:

[0092] Step S301, determining the comprehensive abnormal motion parameter of the device corresponding to different monitoring blocks, or the human abnormal parameter of the human body. If the comprehensive abnormal motion parameter or the human abnormal parameter is greater than or equal to the preset value, the monitoring block is marked as abnormal.

[0093] Step S302, determining the total number of abnormal blocks of the monitoring block with an abnormal mark, and determining the number of chain abnormal blocks of the monitoring block with an abnormal mark that has a chain relationship effect, and calculating the chain abnormal ratio of the number of chain abnormal blocks to the total number of abnormal blocks.

[0094] Step S303, determining the hidden danger warning level of the smelting site based on the comprehensive abnormal motion parameter of different monitoring blocks, or the human abnormal parameter of the human body, and the chain abnormal ratio.

[0095] In some embodiments disclosed in the present invention, the expression for determining the hidden danger warning level of the smelting site is:

[0096] ;

[0097] Wherein, is the hidden danger warning level, is the number of chain abnormal blocks, is the total number of abnormal blocks, is the influence adjustment coefficient of the chain abnormal ratio, is the influence adjustment constant of the chain abnormal ratio, is the comprehensive abnormal motion parameter of the i-th monitoring block with an abnormal mark, is the human abnormal parameter of the i-th monitoring block with an abnormal mark, and n is the total number of monitoring blocks with an abnormal mark.

[0098] Step S400: Determine the hidden danger alarm level based on the hidden danger warning level interval to which the hidden danger warning degree belongs.

[0099] Based on the hidden danger warning degree determined in step S300, compare it with the preset hidden danger warning degree intervals. These intervals are usually determined based on historical data and expert experience, and each interval corresponds to a specific hidden danger alarm level. The level of the hidden danger alarm directly reflects the current safety risk level of the smelting site. Once the hidden danger warning degree falls into a specific interval, the corresponding alarm mechanism is immediately triggered to notify relevant personnel to take measures for disposal in a timely manner. By setting different alarm levels, hierarchical management of the safety hidden dangers in the smelting site can be achieved, improving the efficiency and effectiveness of emergency response.

[0100] In some embodiments disclosed by the present invention, there is also disclosed a smelting hidden danger warning system based on video data analysis, including:

[0101] The first module is used to divide the monitoring blocks for the video images captured by each camera, and determine the monitoring strategy for the monitoring block based on the objects to be monitored in the monitoring block.

[0102] The second module is used to analyze the video images corresponding to each monitoring block, determine the status description of the equipment or the behavior description of the human body corresponding to the monitoring block, and analyze the status description or behavior description using the monitoring strategy corresponding to each monitoring block to determine the abnormal situation assessment of the corresponding equipment or human body.

[0103] The third module is used to construct a floor plan of the smelting site, mark each monitoring block on the floor plan of the smelting site, associate different monitoring blocks based on the chain effect characteristics between equipment, monitor the abnormal situation assessment of each monitoring block in real time, mark the monitoring blocks whose abnormal situation assessment is determined to be abnormal, analyze the chain effect relationship of the retrieved blocks with abnormal marks, and determine the hidden danger warning degree of the smelting site based on the analysis result.

[0104] The fourth module is used to determine the hidden danger alarm level based on the hidden danger warning level interval to which the hidden danger warning degree belongs.

[0105] The present invention discloses a method and system for warning of smelting hazards based on video data analysis, which relates to the technical field of early warning and control in smelting factories. Specifically, it discloses dividing monitoring blocks for the video images captured by cameras and determining monitoring strategies according to monitoring objects; analyzing the video images of the monitoring blocks to identify equipment states or human behaviors, and using the monitoring strategies to evaluate abnormal situations; constructing a floor plan of the smelting site, marking the monitoring blocks, and performing correlation analysis based on the characteristics of equipment interlock effects to monitor abnormal situations in real time, marking the abnormal blocks, and analyzing their interlock effects to determine the degree of hazard warning; determining the corresponding hazard alarm level according to the interval to which the hazard warning degree belongs. The above technical solution of the present invention effectively improves the ability to identify safety hazards and the emergency response speed in the smelting site through refined monitoring, real-time analysis, interlock effect analysis, and determination of alarm levels.

[0106] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by hardware or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various implementation scenarios of the present invention.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A smelting hazard warning method based on video data analysis, characterized in that Including: Dividing the monitored area for the video images captured by each camera, and determining the monitoring strategy for the monitored area based on the objects to be monitored in the monitored area; Analyzing the video images corresponding to each monitored area to determine the state description of the device or the behavior description of the human body corresponding to the monitored area, and using the monitoring strategy corresponding to each monitored area to analyze the state description or the behavior description to determine the corresponding evaluation of the abnormal situation of the device or the human body; Constructing a plan view of the smelting site, marking each monitored area on the plan view of the smelting site, associating different monitored areas with chain effects based on the chain effect characteristics between devices, monitoring the evaluation of the abnormal situation of each monitored area in real time, marking the monitored areas whose evaluation of the abnormal situation is recognized as abnormal, analyzing the chain effect relationship of the retrieved areas with abnormal marks, and determining the degree of potential hazard warning of the smelting site based on the analysis results; Determining the potential hazard alarm level based on the potential hazard warning degree interval to which the potential hazard warning degree belongs; The method for analyzing the chain effect relationship of the retrieved areas with abnormal marks and determining the degree of potential hazard warning of the smelting site based on the analysis results includes: Determining the comprehensive abnormal motion parameters of the devices corresponding to different monitored areas, or the human body abnormal parameters of the human body. If the comprehensive abnormal motion parameters or the human body abnormal parameters are greater than or equal to the preset value, the monitored area is marked as abnormal. Among them, the method for determining the comprehensive abnormal motion parameters includes performing real-time motion analysis on the characteristic edges or characteristic marks in the monitored area to determine the first abnormal motion parameter and the second abnormal motion parameter of the device, and determining the comprehensive abnormal motion parameters of the device corresponding to the monitored area based on the first abnormal motion parameter and the second abnormal motion parameter; Determining the total number of abnormal areas of the monitored areas marked as abnormal, determining the number of chained abnormal areas of the monitored areas marked as abnormal with chain effect influence, and calculating the chained abnormal ratio of the number of chained abnormal areas to the total number of abnormal areas; Determining the degree of potential hazard warning of the smelting site based on the comprehensive abnormal motion parameters of different monitored areas, or the human body abnormal parameters of the human body, and the chained abnormal ratio.

2. The method for warning smelting hazards based on video data analysis according to claim 1, wherein The method for dividing the monitored area in the video image includes: Determining the key equipment area and the personnel operation area in the video image, and dividing the video image based on the key equipment area and the personnel operation area.

3. The method for warning smelting hazards based on video data analysis according to claim 1, wherein, The method for determining the monitoring strategy for the monitored area based on the objects to be monitored in the monitored area includes: Constructing several abnormal state types for the abnormal states of the key equipment area, and for each abnormal state type, constructing several abnormal state levels, and each abnormal state level corresponds to several abnormal state judgment factor parameter intervals; Designating several personnel sub-operation areas in the personnel operation area, constructing several abnormal behavior types for the abnormal human behaviors in each personnel sub-operation area, and for each abnormal behavior type, constructing several abnormal behavior levels, and each abnormal behavior level corresponds to several abnormal behavior judgment factor parameter intervals.

4. The smelting hazard warning method based on video data analysis according to claim 1, characterized in that, A method for analyzing the video image corresponding to each monitoring block and determining the status description of the device corresponding to the monitoring block includes: Delimit the feature edges or mark the feature signs of the key structures of the devices in the monitoring block; Based on the normal operating state of the devices in the monitoring block, determine the movement ranges of different feature edges or feature signs, delimit several levels of sub-movement ranges for the movement ranges, and configure the first abnormal movement parameters for each sub-movement range; Based on the normal operating state of the devices in the monitoring block, determine the movement speed intervals of different feature edges or feature signs, delimit several levels of sub-movement speed intervals for the movement speed intervals, and configure the second abnormal movement parameters for each sub-movement speed interval; Recognize the first abnormal movement parameter and the second abnormal movement parameter as the status description of the devices in the monitoring block.

5. The smelting hazard warning method based on video data analysis according to claim 4, wherein A method for determining the behavior description of the human body in the monitoring block includes: Use human body recognition technology to perform real-time recognition of the human body in the monitoring block, and construct a real-time human body limb expression model for the recognized human body; Determine the different personnel working areas in the monitoring block, determine several normal human postures for each personnel working area, construct a normal human body limb expression model based on the normal human postures, and record the combination of several normal human body limb expression models as the normal human body limb expression model group; Use the normal human body limb expression model group to perform comparison and analysis on the real-time human body limb expression model. If no matching normal human body limb expression model is found in the normal human body limb expression model group, mark the real-time human body limb expression model as abnormal; Count the number of abnormal behaviors of the real-time human body limb expression model marked as abnormal within a preset time period, and recognize it as the behavior description of the human body in the monitoring block based on the number of abnormal behaviors.

6. The method for warning smelting hazards based on video data analysis according to claim 5, characterized in that, A method for analyzing the status description or behavior description using the monitoring strategy corresponding to each monitoring block and determining the abnormal situation assessment of the corresponding device or human body includes: Perform real-time movement analysis on the feature edges or feature signs in the monitoring block, determine the first abnormal movement parameter and the second abnormal movement parameter of the device, and determine the comprehensive abnormal movement parameter of the device corresponding to the monitoring block based on the first abnormal movement parameter and the second abnormal movement parameter; Among them, the expression for calculating the comprehensive abnormal movement parameter is: ； Among them, is the comprehensive abnormal motion parameter, is the influence weight of the first abnormal motion parameter, is the first abnormal motion parameter, is the influence weight of the second abnormal motion parameter, is the second abnormal motion parameter, is the continuous abnormal state duration of the device, is the influence adjustment coefficient of the continuous abnormal state duration, and c is the influence adjustment constant of the continuous abnormal state duration; For the number of abnormal behaviors of the real-time human body limb expression model, several abnormal behavior number intervals are constructed, and a human body abnormal parameter is set for each abnormal behavior number interval. By judging the abnormal behavior number interval to which the number of abnormal behaviors belongs, the human body abnormal parameter of the real-time human body limb expression model is determined.

7. The smelting hazard warning method based on video data analysis according to claim 1, characterized in that, The expression for determining the hidden danger warning level of the smelting site is: ； Among them, is the hidden danger warning level, is the number of chained abnormal blocks, is the total number of abnormal blocks, is the influence adjustment coefficient of the chained abnormal ratio, is the influence adjustment constant of the chained abnormal ratio, is the comprehensive abnormal motion parameter of the i-th abnormal marker monitoring block, is the human body abnormal parameter of the i-th abnormal marker monitoring block, and n is the total number of abnormal marker monitoring blocks.

8. A smelting hazard warning system based on video data analysis, characterized in that, A smelting hidden danger warning method for executing any one of claims 1-7 includes: A first module for dividing the monitoring blocks for the video images captured by each camera and determining the monitoring strategy of the monitoring block based on the objects to be monitored in the monitoring block; The second module is used to analyze the video images corresponding to each monitoring block, determine the status description of the device or the behavior description of the human body corresponding to the monitoring block, and use the monitoring strategy corresponding to each monitoring block to analyze the status description or behavior description to determine the corresponding abnormal situation assessment of the device or the human body; The third module is used to construct a plan view of the smelting site, mark each monitoring block on the plan view of the smelting site, based on the interlocking influence characteristics between devices, associate different monitoring blocks with interlocking influences, monitor the abnormal situation assessment of each monitoring block in real time, mark the monitoring blocks whose abnormal situation assessment is determined to be abnormal, analyze the interlocking influence relationship of the retrieved blocks with abnormal marks, and based on the analysis results, determine the hidden danger warning level of the smelting site; The fourth module is used to determine the hidden danger alarm level based on the hidden danger warning level interval to which the hidden danger warning level belongs.

Images