Unmanned chemical plant monitoring system based on the Internet of Things

Abstract

The present invention discloses an unmanned chemical factory monitoring system based on the Internet of Things, which relates to the field of factory supervision technology, and includes: a visual module, a smoke detection module and a supervision device; the visual module includes a camera 1 and a camera 2, the camera 1 is multiple and monitors one or more pre-divided monitoring areas respectively, and the monitoring area of ​​the camera 2 covers the monitoring areas of multiple cameras 1; the supervision device includes a front-end AI unit and a supervision background connected to the front-end AI unit, and the front-end AI unit is integrated in the production equipment with visual function in the unmanned chemical factory. The present application has the effect of reducing the interference of structural facilities in the unmanned chemical factory on safety monitoring.

Description

Technical Field

[0001] The present application relates to the technical field of factory supervision, and in particular to an unmanned factory monitoring system based on the Internet of Things. Background Art

[0002] An unmanned factory means that raw materials are input from one end of the factory, and under predefined production instructions, they go through product design, process design, production and processing, inspection and packaging, and finally the products are output from the other end of the factory. The work is done by computer-controlled robots, CNC machine tools, unmanned transport vehicles and automated warehouses, and workers do not directly participate in the work.

[0003] In this highly automated production environment, safety issues are particularly important. Fire is one of the most serious accidents in industrial production. Currently, in unmanned factories, the factory environment is monitored by installing high-definition cameras, sensors and other equipment. However, in a particularly complex factory environment, due to the cross-operation of various automated equipment, the monitoring line of sight will be blocked, resulting in inadequate monitoring and easy omissions. Therefore, this application proposes a new technical solution. Summary of the invention

[0004] In order to reduce the interference of structural facilities in unmanned factories with safety monitoring, the present application provides an unmanned factory monitoring system based on the Internet of Things.

[0005] This application provides an unmanned chemical plant monitoring system based on the Internet of Things, which adopts the following technical solutions:

[0006] An unmanned chemical plant monitoring system based on the Internet of Things, comprising:

[0007] A visual module is arranged in the workshop and its visual perception range at least covers the working area of ​​the workshop;

[0008] a smoke detection module, which is deployed in the workshop and performs smoke detection; and,

[0009] A monitoring device electrically connected to the visual module, the smoke detection module and the unmanned chemical plant production control system;

[0010] The visual module includes camera 1 and camera 2. The camera 1 is multiple and monitors one or more pre-divided monitoring areas respectively. The monitoring area of ​​camera 2 covers the monitoring areas of multiple camera 1s. The monitoring device includes a front-end AI unit and a monitoring background connected to the front-end AI unit. The front-end AI unit is integrated into a production equipment with visual functions in an unmanned factory. The front-end AI unit is configured as follows:

[0011] Obtain image data collected during production and identify suspected fires; and,

[0012] If suspected fire features are identified, the preset warning action will be executed, and the image with the features located will be uploaded to the supervision backend;

[0013] The supervisory backend configuration is:

[0014] Define the monitoring data fed back by camera 1 and camera 2 as data 1 and data 2 respectively;

[0015] If an image is received from the front-end AI unit, the suspected fire location is located, and the database is searched to determine the matching camera 1, camera 2, and smoke detection module, and data 1, data 2, and smoke detection information are obtained as reference information;

[0016] Fire diagnosis is performed based on images and reference information fed back by the front-end AI unit;

[0017] Based on the fire diagnosis results, the front-end AI unit will cancel the alarm or execute the preset fire response action.

[0018] Optionally, at least the camera 1 is installed on a mobile platform, the mobile platform is electrically connected to a supervision background, and the supervision background is configured as follows:

[0019] Record the production process corresponding to the monitoring area of ​​camera 1, and record the production equipment corresponding to the production process;

[0020] Based on the suspected fire location and the location of the matching camera 1, the production process in the middle is determined;

[0021] The real-time operation behavior of the production process in the middle is obtained through the unmanned production control system;

[0022] Conduct monitoring line of sight obstruction diagnosis and processing based on the suspected fire location, the location of camera 1, and real-time operation behavior;

[0023] If the diagnosis result meets the preset occlusion condition, the mobile platform is controlled to adjust the position of the camera 1.

[0024] Optionally, the supervision background is configured as follows:

[0025] Obtain the real-time operation behavior of the production process in the middle and the operation trajectory that changes with the time axis;

[0026] If the diagnosis result meets the preset obstruction condition, the obstruction duration t2 is determined based on the operation trajectory to determine whether it is greater than the preset threshold value three. If so, the mobile platform is controlled to adjust the position of camera one; if not, the data one obtained after duration t2 is called for fire diagnosis, and the mobile platform remains stationary.

[0027] Optionally, the supervision background is configured as follows:

[0028] Look up the preset fire impact level table based on the suspected fire location;

[0029] If the real-time fire level exceeds the preset threshold value one, the unmanned production control system is used to obtain the preceding and following operation behaviors corresponding to the preceding and following production processes of the suspected fire location;

[0030] The time t1 required for the pre- and post-assessment operations to negotiate with the suspected fire location;

[0031] If the duration t1 is less than the preset threshold value 2, the production line speed reduction control action is executed, and the real-time operation behavior for monitoring the line of sight obstruction diagnosis processing is updated.

[0032] Optionally, the supervision background is configured as follows:

[0033] Defining fire identification interference behaviors in production; wherein the fire identification interference behaviors include welding behaviors and metal cutting behaviors;

[0034] If there is a fire identification interference behavior at the suspected fire feature position, determine whether the suspected fire is a safety feature caused by the fire identification interference behavior, and if so, terminate the fire diagnosis;

[0035] Determine whether the real-time operation behavior of the intermediate production process has any fire identification interference behavior. If so, call data one when terminating the fire diagnosis.

[0036] Optionally, the supervision background is configured as follows:

[0037] Splicing the surveillance range contours of multiple cameras to obtain a surveillance contour splicing image;

[0038] If a camera adjusts its viewing angle, the monitoring contour mosaic image is updated and compared with the monitoring contour mosaic image before adjusting the viewing angle to obtain the missing contour;

[0039] Adjust other nearby camera pairs to compensate for the missing outline.

[0040] Optionally, an acceleration sensor is fixed to the moving part of the mobile platform, and the acceleration sensor is connected to the supervision background. The supervision background is configured as follows: if the movement of the mobile platform is controlled and the detection information fed back by the acceleration sensor meets the preset static condition, a corresponding mobile platform fault prompt is output, and the data obtained after the time t2 is called for fire diagnosis.

[0041] Optionally, a backup battery pack is also included for powering the visual module and the smoke detection module.

[0042] In summary, this application includes the following beneficial technical effects:

[0043] 1. The visual function of the production equipment in the unmanned factory can be used to monitor whether there are any suspected fire characteristics in the workshop. Because the visual monitoring is carried out along with the production equipment, the probability of being blocked by other facilities when monitoring a specific location is smaller. At the same time, compared with visual perception from above the workshop or far away from the working position, there is less interference.

[0044] 2. Suspected fire characteristics will be reported to the supervision backend for further identification. Therefore, on the one hand, the computing power requirements for the front end are lower, and on the other hand, the backend receives less data and is more targeted; at the same time, the accuracy of further identification is higher;

[0045] 3. The supervision background also combines the workshop area monitoring camera 1, the camera 2 with a larger monitoring range, and the smoke detection information of the smoke detection module to further diagnose the fire. Therefore, it can also make up for the blank areas caused by the visual monitoring of production equipment, and can monitor the fire from multiple angles and dimensions to reduce the chance of missed reports and false reports. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic diagram of the overall architecture of this application. DETAILED DESCRIPTION

[0047] The following is combined with Figure 1 This application is described in further detail.

[0048] The embodiment of the present application discloses an unmanned chemical factory monitoring system based on the Internet of Things.

[0049] Reference Figure 1 The unmanned chemical factory monitoring system based on the Internet of Things includes a visual module, a smoke detection module and a monitoring device, wherein the visual module includes camera 1 and camera 2, the camera 1 is multiple and monitors one or more pre-divided monitoring areas respectively, and the monitoring area of ​​camera 2 covers multiple monitoring areas of camera 1; camera 1 and camera 2 are respectively installed in the workshop through columns and brackets, and the visual perception range at least covers the working area of ​​the workshop; camera 1 and camera 2 can be zoom gun cameras, ball cameras, and hemispherical cameras, especially camera 1, which should be a camera with electrically adjustable viewing angle, that is, the selected camera can be rotated and zoomed to meet diverse usage needs.

[0050] The smoke detection module includes multiple smoke detectors, which are arranged at various locations in the workshop for smoke detection and alarm. It can be understood that each smoke detector has a corresponding optimal detection area, and the combination of each detection area should at least cover the working area of ​​the workshop.

[0051] The monitoring device includes a front-end AI unit and a monitoring background. The front-end AI unit refers to a chip loaded with an AI image recognition program, which is integrated on production equipment with visual functions in an unmanned factory or obtained by loading the AI ​​image recognition program on the control chip of the production equipment. The AI ​​image recognition in this embodiment is to identify flames. It should be noted that the image recognition here does not have high requirements for accuracy, and it is sufficient to detect bright spots that are obviously abnormal from other structures. It is mainly to reduce the uploading of a large amount of non-suspicion data.

[0052] The supervision background is composed of a server or cloud server in the factory and a computer in the workshop. The supervision background is connected to the visual module, smoke detection module, and front-end AI unit in wireless and wired ways, and is connected to the unmanned factory production control system to obtain various data for security supervision and response.

[0053] In this embodiment, the front-end AI unit is configured as follows:

[0054] Obtain image data collected during production and identify suspected fires; and,

[0055] If suspected fire characteristics are identified, the preset warning action will be executed, and the image where the characteristics are located will be uploaded to the supervision background.

[0056] For example, the above-mentioned suspected fire identification can be performed by performing image recognition and identifying bright blocks as suspected fire features; the warning action can be lighting up several preset yellow lights on the device, etc.

[0057] The supervision backend configuration is:

[0058] S11. Define the monitoring data fed back by camera 1 and camera 2 as data 1 and data 2 respectively.

[0059] S12. If an image is fed back by the front-end AI unit, the suspected fire location is located. It can be understood that each device has an ID and a location, so the approximate location of the fire can be determined by who the image comes from. Furthermore, the features in the image can be further located, combined with the location of the device, to obtain a relatively accurate suspected fire location.

[0060] Afterwards, the database is searched to determine the matching camera 1, camera 2 and smoke detection module, and data 1, data 2 and smoke detection information are obtained as reference information.

[0061] It has been explained above that each camera 1 and camera 2 has its own monitoring area. If the suspected fire location is within a certain monitoring area, it can be determined which camera 1 and camera 2 to match; the smoke detection module can be determined in the same way.

[0062] It should be noted that Camera 2 is designed for full-area monitoring, or large-scale monitoring of multiple monitoring areas. On the one hand, it fills in the blind spots of Camera 1 during use, and on the other hand, it can also serve as the basis for overall security supervision, such as anti-theft and personnel entry and exit supervision.

[0063] S13. Perform fire diagnosis based on the images and reference information fed back by the front-end AI unit.

[0064] Example: First, the more abundant computing power provided by the server is used to perform more detailed image recognition on the feedback images, that is, it is processed with a more accurate image recognition model to identify and verify the flame characteristics; secondly, the real-time images of data one and data two are extracted for image recognition to identify and determine whether there are flames and smoke characteristics; finally, it is determined whether the smoke detection information meets the characteristics of a fire (signal characteristics); any of the above determines the presence of smoke and flames, that is, the presence of a fire is diagnosed; if there is a conflict between any two of the above results, such as the feedback image recognizes a predefined observable medium-sized flame, but the recognition results of data one and data two are that it does not exist, it is diagnosed as a false alarm; the above achieves comprehensive supervision of small fires, medium and large fires, and fires without open flames.

[0065] S14. Instruct the front-end AI unit to cancel the alarm or execute a preset fire response action based on the fire diagnosis result; the fire response action may be turning on the alarm lights of the workshop observation window and the control console, and a computer pop-up window showing an image corresponding to the fire location.

[0066] Based on the above content, we can know that this system:

[0067] 1. The visual function of the production equipment in the unmanned factory can be used to monitor whether there are any suspected fire characteristics in the workshop. Because the visual monitoring is carried out along with the production equipment, the probability of being blocked by other facilities when monitoring a specific location is smaller. At the same time, compared with visual perception from above the workshop or far away from the working position, there is less interference.

[0068] 2. Suspected fire characteristics will be reported to the supervision backend for further identification. Therefore, on the one hand, the computing power requirements for the front end are lower, and on the other hand, the backend receives less data and is more targeted; at the same time, the accuracy of further identification is higher;

[0069] 3. The supervision background also combines the workshop area monitoring camera 1, the camera 2 with a larger monitoring range, and the smoke detection information of the smoke detection module to further diagnose the fire. Therefore, it can also make up for the blank areas caused by the visual monitoring of production equipment, and can monitor the fire from multiple angles and dimensions to reduce the chance of missed reports and false reports.

[0070] In this embodiment, in order to further reduce the interference caused by the obstruction of the fire position during the operation of the production equipment, at least one camera is installed on a mobile platform, which can be a small cross slide, and the camera is installed on the slider of the cross slide. The supervision background configuration is:

[0071] S21. Record the production process corresponding to the monitoring area of ​​camera 1, and record the production equipment corresponding to the production process.

[0072] Example: Assign a camera 1 to each production link, and establish a one-to-one correspondence between camera 1, production link, and production equipment.

[0073] S22. Determine the production process in the middle based on the suspected fire location and the location of the matching camera 1.

[0074] It is known that two points determine a straight line, the suspected fire location is point one and the location of camera one is point two, and the production process passed by the line connecting the two points is the production process in the middle.

[0075] S23. The real-time operation behavior of the production process in the middle is obtained through the unmanned production control system. It can be understood that each action in the unmanned production process is executed according to the instructions issued by the control system, and the action is a closed loop, that is, there will be feedback after the action. Therefore, the feedback and control instructions of various equipment can be obtained from the unmanned production control system to identify the real-time operation behavior. For example: the AGC car transfers the semi-finished product from area A to area B, and the transfer behavior is obtained according to the feedback of the AGV car, and its location is determined according to the relevant parameters and time.

[0076] S24. Perform monitoring line of sight obstruction diagnosis processing based on the suspected fire location, the position of camera one and real-time operation behavior; example: mark points / generate outlines for each location in the three-dimensional space coordinate system, and a real-time operating device (or a workpiece on the device) is located between the fire location and the position of camera one.

[0077] S25. If the diagnosis result meets the preset obstruction condition (such as: the above three positions are in line, the outline of the equipment / workpiece obstructs the suspected fire position), the mobile platform is controlled to adjust the position of camera 1. For example: the cross slide is installed vertically. At this time, the camera is moved up and down and horizontally until it is unobstructed. The up and down and horizontal movement amounts are in accordance with the preset movement rules. Note that after the fire diagnosis is completed, the cross slide needs to be reset to avoid disrupting the workshop monitoring layout.

[0078] According to the above content, this system will analyze whether Camera 1 is blocked based on the production behavior in the workshop, and determine the position of Camera 1 based on the result to reduce the interference of production behavior on visual monitoring.

[0079] Furthermore, the supervision backend is also configured as follows:

[0080] Obtain the operation trajectory of the real-time operation behavior of the production process in the middle that changes with the time axis; Example: Generate a continuous contour position change graph based on the time axis and call the historical behavior monitoring video;

[0081] If the diagnosis result meets the preset occlusion condition, the occlusion duration t2 is determined based on the operation trajectory to determine whether it is greater than the preset threshold value three (e.g., 3s). If so, the mobile platform is controlled to adjust the position of camera one; if not, the data one obtained after duration t2 is called for fire diagnosis, and the mobile platform remains stationary.

[0082] According to the above content, if the operation behavior that causes occlusion passes relatively quickly, the position of Camera 1 will not be adjusted, but will wait to avoid the situation where the position of Camera 1 moves and then the original position is not obstructed. At the same time, it can be concluded here that the movement of Camera 1 should be opposite to the direction of the operation trajectory change, to avoid moving for 2s and then moving to the front or back of the new position of the device 2s later.

[0083] On the basis of the above, the system installs an acceleration sensor on the moving part of the mobile platform (i.e. the slider of the cross slide). The acceleration sensor and the corresponding mobile platform can share a wireless data transmission terminal to feed back the signal to the supervision background; the supervision background configuration is:

[0084] If the mobile platform is controlled to move and the detection information fed back by the acceleration sensor meets the preset static condition (such as acceleration is 0), the corresponding mobile platform fault prompt is output, and the data obtained after time t2 is called for fire diagnosis.

[0085] According to the above settings, this system can detect mobile platform failures in a timely manner to reduce misjudgments caused by mobile platform failures.

[0086] In another embodiment of the system, the supervision background is configured as follows:

[0087] S31. Look up the preset fire impact level table according to the suspected fire location. The table records the fire impact levels corresponding to flames and smoke of various sizes, such as: level one - dangerous fire at a single workstation, level two - dangerous fire at a single equipment, level three - dangerous fire at a single production link, and dangerous fire at multiple links.

[0088] S32. If the real-time fire level exceeds a preset threshold level one (e.g., level two), the unmanned production control system is used to obtain the preceding and following operation behaviors corresponding to the preceding and following production processes of the suspected fire location;

[0089] Evaluate the time t1 required for the previous and subsequent work behaviors to interact with the suspected fire location. For example: obtain the time of the next action at the designated fire location based on the historical monitoring video and the current time to calculate the time t1; obtain the time of the next workpiece moving to the fire location based on the historical monitoring video and the current time to calculate the time t1.

[0090] S33. If the duration t1 is less than the preset threshold value 2, the production line speed reduction control action is executed (for example, the original production rate is defined as 1, and now it is reduced to 0.8), and the real-time operation behavior used to monitor the line of sight occlusion diagnosis processing is updated.

[0091] According to the above content, this system will not handle fires of different sizes in a unified manner. If the fire is relatively large, it will be combined with production behavior analysis and, if necessary, slow down production to avoid a suspected large fire in the next link while new workpieces continue to be delivered in the previous link. In this way, this system can reduce economic losses after a fire. After all, apart from the workpieces, many times a fire occurs because the equipment is damaged. If the previous equipment is damaged and shuts down or even moves randomly, and the next equipment (robotic arm) follows, it is likely to cause equipment collision and cause significant economic losses.

[0092] In this system, considering the welding and metal cutting in unmanned workshops such as automobile assembly, which will produce flying sparks, the supervision background configuration is as follows:

[0093] Define the fire identification interference behaviors in production; the fire identification interference behaviors include welding behaviors and metal cutting behaviors;

[0094] If there is fire identification interference at the suspected fire feature location (the production link), determine whether the suspected fire is a safety feature caused by the fire identification interference (i.e., whether the recognition image is a spark or flame of cutting or welding). If so, terminate the fire diagnosis;

[0095] Determine whether the real-time operation behavior of the intermediate production process has any fire identification interference behavior. If so, call data one when terminating the fire diagnosis.

[0096] Based on the above content, it can be seen that this system will also identify the interference caused by welding and metal cutting in the workshop's production activities, and adjust the fire diagnosis process in a timely manner to reduce the occurrence of misjudgment.

[0097] In another embodiment of the system, the supervision background is configured as follows:

[0098] Splicing the surveillance range contours of multiple cameras to obtain a surveillance contour splicing image;

[0099] If a camera adjusts its viewing angle, the monitoring contour mosaic image is updated and compared with the monitoring contour mosaic image before adjusting the viewing angle to obtain the missing contour;

[0100] Adjust other nearby cameras to compensate for the missing outline. Compensation examples: The first method: adjust the focal length, and let another camera one adjacent to the monitoring range outline reduce the focal length. The shorter the focal length, the larger the monitoring range is, so as to compensate; the second method: rotate other cameras in sequence, such as 3 horizontally distributed cameras, one camera moves 10 cm to the left, and the other two cameras move 10 cm to the left synchronously; if it is the second method, there should be an area in the original monitoring range of at least one of the outermost cameras that does not need to be monitored; this embodiment preferably uses the first method for compensation; that is, camera one is a variable-focus camera.

[0101] In one embodiment of the present system, in order to avoid power outages in multiple places in the workshop due to fire and cause system failure, a backup battery pack is also included, which is used to power at least the visual module and the smoke detection module. Example: The conventional power supply network leads to a coil of a series relay, and the normally closed contact of the relay is connected in series with the backup battery pack and the power supply circuit of the visual module and the smoke detection module; when the conventional power supply network is powered off, the normally closed contact of the relay returns to the closed state to conduct the visual module and the smoke detection module, ensuring that the workshop has at least a certain monitoring capability.

[0102] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. An unmanned chemical plant monitoring system based on the Internet of Things, characterized in that: include: A visual module is arranged in the workshop and its visual perception range at least covers the working area of ​​the workshop; A smoke detection module, which is deployed in the workshop and performs smoke detection; and ， A monitoring device electrically connected to the visual module, the smoke detection module and the unmanned chemical plant production control system; The visual module includes camera 1 and camera 2. The camera 1 is multiple and monitors one or more pre-divided monitoring areas respectively. The monitoring area of ​​camera 2 covers the monitoring areas of multiple camera 1s. The monitoring device includes a front-end AI unit and a monitoring background connected to the front-end AI unit. The front-end AI unit is integrated into a production equipment with visual functions in an unmanned factory. The front-end AI unit is configured as follows: Obtain image data collected during production and identify suspected fires; and, If suspected fire features are identified, the preset warning action will be executed, and the image with the features located will be uploaded to the supervision backend; The supervisory backend configuration is: Define the monitoring data fed back by camera 1 and camera 2 as data 1 and data 2 respectively; If an image is received from the front-end AI unit, the suspected fire location is located, and the database is searched to determine the matching camera 1, camera 2, and smoke detection module, and data 1, data 2, and smoke detection information are obtained as reference information; Fire diagnosis is performed based on images and reference information fed back by the front-end AI unit; According to the fire diagnosis results, the front-end AI unit can cancel the alarm or execute the preset fire response action; At least the camera 1 is installed on a mobile platform, the mobile platform is electrically connected to a supervision background, and the supervision background is configured to: record the production process corresponding to the monitoring area of ​​the camera 1, and record the production equipment corresponding to the production process; Based on the suspected fire location and the location of the matching camera 1, the production process in the middle is determined; The real-time operation behavior of the production process in the middle is obtained through the unmanned production control system; Conduct monitoring line of sight obstruction diagnosis and processing based on the suspected fire location, the location of camera 1, and real-time operation behavior; If the diagnosis result meets the preset occlusion condition, the mobile platform is controlled to adjust the position of the camera 1.

2. The unmanned chemical plant monitoring system based on the Internet of Things according to claim 1 is characterized in that: The supervisory backend configuration is: Obtain the real-time operation behavior of the production process in the middle and the operation trajectory that changes with the time axis; If the diagnosis result meets the preset obstruction condition, the obstruction duration t2 is determined based on the operation trajectory to determine whether it is greater than the preset threshold value three. If so, the mobile platform is controlled to adjust the position of camera one; if not, the data one obtained after duration t2 is called for fire diagnosis, and the mobile platform remains stationary.

3. The unmanned chemical plant monitoring system based on the Internet of Things according to claim 1 is characterized in that: The supervisory backend configuration is: Look up the preset fire impact level table based on the suspected fire location; If the real-time fire level exceeds the preset threshold value one, the unmanned production control system is used to obtain the preceding and following operation behaviors corresponding to the preceding and following production processes of the suspected fire location; The time t1 required for the pre- and post-assessment operations to negotiate with the suspected fire location; If the duration t1 is less than the preset threshold value 2, the production line speed reduction control action is executed, and the real-time operation behavior for monitoring the line of sight obstruction diagnosis processing is updated.

4. The unmanned chemical plant monitoring system based on the Internet of Things according to claim 1 is characterized in that: The supervisory backend configuration is: Defining fire identification interference behaviors in production; wherein the fire identification interference behaviors include welding behaviors and metal cutting behaviors; If there is a fire identification interference behavior at the suspected fire feature position, determine whether the suspected fire is a safety feature caused by the fire identification interference behavior, and if so, terminate the fire diagnosis; Determine whether the real-time operation behavior of the intermediate production process has any fire identification interference behavior. If so, call data one when terminating the fire diagnosis.

5. The unmanned chemical plant monitoring system based on the Internet of Things according to claim 1 is characterized in that: The supervisory backend configuration is: Splicing the surveillance range outlines of multiple cameras to obtain a surveillance outline splicing map; If a camera adjusts its viewing angle, the monitoring contour mosaic image is updated and compared with the monitoring contour mosaic image before adjusting the viewing angle to obtain the missing contour; Adjust other nearby camera pairs to compensate for the missing outline.

6. The unmanned chemical factory monitoring system based on the Internet of Things according to claim 2 is characterized in that: An acceleration sensor is fixed to the moving part of the mobile platform, and the acceleration sensor is connected to the supervision background. The supervision background is configured as follows: if the mobile platform is controlled to move and the detection information fed back by the acceleration sensor meets the preset static condition, the corresponding mobile platform fault prompt is output, and the data obtained after the time length t2 is called for fire diagnosis.

7. The unmanned chemical plant monitoring system based on the Internet of Things according to claim 1 is characterized in that: It also includes a backup battery pack for powering the vision module and smoke detection module.

Images