Control panel AI monitoring method

Abstract

The invention discloses a control panel AI monitoring method, which belongs to the field of intelligent monitoring, and comprises the following steps: dividing a panel identification area, uploading a stereogram and a preset front view of a panel, and carrying out component labeling; setting shooting intervals and times; setting a datum point, a datum line or a graph; creating an alarm list, and setting values and times; the front view is called for comparison after recognition of the recognition area; if yes, giving an alarm, judging whether a view is converted and judging new and old converted data, and setting the new converted data as default; if not, judging whether the view is a conversion view, if so, recording conversion failure and performing conversion again, and otherwise, performing conversion by default; when no data exists or conversion is carried out again, the position of the stereogram is adjusted to be matched with the photograph, data obtained by converting the stereogram into a front view is reserved as conversion data, and comparison is carried out again; comparing with an alarm value; panel states at different angles can be automatically identified and judged, data are automatically adjusted and converted, 24-hour rapid and accurate inspection is achieved, panel transformation is avoided, and the applicability is high.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of intelligent monitoring, and particularly relates to an AI monitoring method for a control panel. BACKGROUND

[0002] There are various types of traditional control panels in a factory, which bear the state control and real-time monitoring functions of various system modules such as power supply, production line transmission, temperature control system, material conveying, etc. The running stability of such panels is directly related to the production continuity of the factory. Once an individual module appears abnormal, such as indicator light failure, instrument value overrun, button jamming, etc., if it cannot be found in time, it may lead to production line shutdown, product quality defects or even safety accidents. Therefore, real-time monitoring of the control panel is crucial.

[0003] When the prior art is used, the traditional monitoring method relies on manual time-fixed point inspection to collect data. This mode has three core problems: first, the labor cost is high. Large factories need to configure full-time inspection personnel, and need to cover multiple shifts and multiple areas, so the long-term operating cost is significant. Second, the detection frequency is low. Manual inspection usually has an interval of 1-2 hours, which is difficult to realize minute-level or second-level real-time monitoring, and the "window period" is easy to appear after the abnormality occurs. Third, the result is subjective. The experience, fatigue degree and attention concentration of the inspection personnel will directly affect the detection accuracy, and it is easy to appear missed inspection and misjudgment, such as weak indicator light failure being ignored, instrument value reading deviation, etc. SUMMARY

[0004] (I) Technical problems to be solved

[0005] In order to overcome the deficiencies of the prior art, the present application proposes an AI monitoring method for a control panel to solve the three core problems of the traditional monitoring method relying on manual time-fixed point inspection to collect data: first, the labor cost is high. Large factories need to configure full-time inspection personnel, and need to cover multiple shifts and multiple areas, so the long-term operating cost is significant. Second, the detection frequency is low. Manual inspection usually has an interval of 1-2 hours, which is difficult to realize minute-level or second-level real-time monitoring, and the "window period" is easy to appear after the abnormality occurs. Third, the result is subjective. The experience, fatigue degree and attention concentration of the inspection personnel will directly affect the detection accuracy, and it is easy to appear missed inspection and misjudgment, such as weak indicator light failure being ignored, instrument value reading deviation, etc.

[0006] Second, to solve the situation when the prior art is used.

[0007] (II) Technical solutions

[0008] The application is implemented by the following technical solutions: the application proposes an AI monitoring method for a control panel, which comprises the following steps:

[0009] S1: According to the monitoring picture of the shooting device, the panel recognition area is divided, and according to the panel type of the panel recognition area, the stereogram and the preset front view of the corresponding panel are uploaded, and the component area of the panel in the preset front view of each area is marked;

[0010] S2: Set the shooting interval and the number of shots;

[0011] S3: Set the reference point for panel position matching, and set one or more reference lines or graphics for panel matching, and set the error value;

[0012] S4: Create an alarm list including communication alarms and on-site alarms, and set the alarm value of each component alarm state, and set the number of conversion failure alarms;

[0013] S5: Recognize the recognition area of the photographed photo, and compare the preset front view of the corresponding recognition area panel with the photo. When comparing, first match the panel position through the reference point, and then compare the reference line or graphic;

[0014] S501: If the comparison result matches, the component alarm judgment is performed, and whether it is a conversion view is judged. If not, ignore it. If yes, judge the new and old conversion data. If it is new conversion data, save the conversion data as default conversion data. If it is old conversion data, do not change it;

[0015] S502: If the comparison result does not match, it is judged whether it is a conversion view. If yes, record the number of conversion failures once and reconvert the view. If not, call the default conversion data for conversion;

[0016] S503: If there is no conversion data or the photo is reconverted, the stereogram of the corresponding panel is called, the corresponding points in the photo and the stereogram are matched according to the reference point, the position of the stereogram is adjusted according to the reference line or graphic, the panel view angle of the stereogram is matched and fitted with the photographed photo, and then the conversion data of the photo is obtained according to the data of the stereogram converted to the front view, and the conversion data is saved. The converted photo is reconverted by S5 reference comparison;

[0017] S6: Compare the photos that have been successfully compared with the alarm value. If the alarm value is reached and the number of conversion failure alarms is reached, alarm according to the alarm list.

[0018] Further, S504 is further provided after S503: Set the automatic cover time or number of saved photos, and mark and pack the photos before and after conversion with time stamp.

[0019] Further, S0 is provided before S1: Set the ordinary account and the administrator account. The ordinary account can call and view the photos, receive and control the alarms, and the administrator account has all the permissions.

[0020] Furthermore, the alarm value in S4 includes one or more combinations of displayed content, brightness, or color. In S6, false alarm and confirmed alarm functions are set. When an alarm occurs, the alarm type is marked as false alarm or confirmed alarm by the function. When a false alarm is triggered within a time period or number of recognitions, the alarm value at the time of the false alarm is recorded as the false alarm value, and the false alarm value with the maximum difference from the alarm value is automatically taken as the new alarm value. When no alarm is triggered within a time period or number of recognitions, the alarm value at the time of the correct alarm is recorded as the correct value, and the midpoint between the correct value with the minimum difference from the alarm value and the alarm value is automatically taken as the new alarm value.

[0021] Furthermore, the S6 also includes a final endpoint function, which is used to limit the alarm value endpoint by correcting the correct value. When the alarm value corrected by the correct value reaches the final endpoint, a fault alarm is triggered.

[0022] Furthermore, prior to S1, there is S001: a camera and an AI monitoring unit are set up in the panel area. The camera is connected to the AI ​​monitoring unit via wired or wireless means, and the AI ​​monitoring unit communicates with the mobile terminal via a PTZ camera.

[0023] Furthermore, before S1, there is S002: setting up an inspection track according to the panel monitoring area, and mounting one or more shooting devices on the inspection track via a track trolley;

[0024] The S4 also sets the fixed stopping position and stopping time of the track trolley at the designated panel.

[0025] Furthermore, before S1, there is S003: an angle adjustment track parallel to the front of the panel is set on the side of the inspection track adjacent to the front of each panel. The angle adjustment track is used for the movement of the track trolley when the camera changes the camera angle. Multiple fixed stopping positions for taking pictures are set on the angle track.

[0026] After S6, there is S7: When the alarm value is triggered, the track trolley drives the shooting equipment to take pictures from different angles on the front of the panel for confirmation.

[0027] Furthermore, the S7 also includes an alarm self-recognition function. This function matches multi-angle recognition photos. When two or more photos have the same component that reaches the alarm value, an alarm is confirmed. If only one photo triggers the alarm value, it is marked as an error in the shooting angle recognition. The erroneous photo is then marked and the subsequent default shooting position is automatically changed to the remaining position on the angle adjustment track.

[0028] Furthermore, after S7, there is also S701: the erroneous photo is sent to the administrator account, the administrator account has the function of correcting the erroneous photo to the correct recognition, when the erroneous photo is corrected to the correct recognition, the photos taken from other angles are marked as having an incorrect shooting angle, the erroneous photo is marked, and the subsequent shooting position is automatically changed back to the correct recognition position.

[0029] (III) Beneficial Effects

[0030] One of the above technical solutions has the following advantages or beneficial effects:

[0031] The panel area is photographed using a camera, and the image is divided into recognition zones to facilitate panel location identification after fixed-point shooting. A 3D image and a preset front view of the corresponding panel are also provided. Component information and alarm values ​​for component alarm detection are marked on the front view. The image is positioned relative to the preset front view by capturing the panel's reference point in the photograph. A match is checked; if a match is found, a view conversion is performed, and the old and new conversion data are compared. The new conversion data is used as the default conversion data. For mismatched conversion views, the number of failures is recorded and the conversion is repeated. For mismatched original images, the default conversion data is used for conversion. For re-conversions and for images without data, a 3D image and photograph are matched using a reference point. First, the 3D image is transformed to match the photograph, and then the front view is converted from the 3D image. The system converts viewpoint data into photographs, compares them with preset front views, and automatically judges based on alarm values ​​to automatically identify panel areas. It can convert and identify panels from different angles using 3D simulation conversion, increasing the accuracy of automatic panel component judgment. It can also automatically learn and adjust default conversion data to increase conversion efficiency. Conversion failures are recorded, and alarms are triggered based on the number of failures for timely adjustments. This enables rapid and accurate 24-hour inspection, speeding up response time, eliminating the need to modify existing panels, and is easy to install with strong applicability. It can also accurately trace faults through photographs, facilitating maintenance, eliminating manual inspection, reducing labor costs, and avoiding fatigue-related misjudgments. It can achieve minute-level or even second-level identification and judgment. Attached Figure Description

[0032] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0033] Figure 1 This is a flowchart illustrating the setting method in Embodiment 1 of the present invention;

[0034] Figure 2 This is a flowchart illustrating the operation method in Embodiment 1 of the present invention;

[0035] Figure 3 This is a schematic diagram of the account setup process in Embodiment 1 of the present invention;

[0036] Figure 4 This is a schematic diagram of the overlay setting process in Embodiment 1 of the present invention;

[0037] Figure 5 This is a flowchart illustrating the operation method in Embodiment 2 of the present invention;

[0038] Figure 6 This is a schematic diagram of the fault diagnosis process in Embodiment 2 of the present invention;

[0039] Figure 7 This is a schematic diagram of the arrangement method in Embodiment 3 of the present invention;

[0040] Figure 8 This is a schematic diagram of the arrangement method in Embodiment 4 of the present invention;

[0041] Figure 9 This is a schematic diagram of the flow structure of the self-core function in Embodiment 5 of the present invention;

[0042] In the picture: Shooting equipment-1, Track trolley-2, Panel-3, Inspection track-4, Angle adjustment track-5. Detailed Implementation

[0043] The present invention will be further described in detail below with reference to embodiments, but the implementation of the present invention is not limited thereto.

[0044] Example 1:

[0045] This invention provides an AI monitoring method for control panels, comprising the following steps:

[0046] S1: Divide the panel 3 recognition area according to the monitoring screen of the shooting device 1, and upload the corresponding panel 3 stereoscopic image and preset front view according to the panel 3 type of the panel 3 recognition area, and then label the component area of ​​the panel 3 in the preset front view of each area.

[0047] S2: Set the photo interval and number of shots;

[0048] S3: Set the reference point for panel 3 position matching, and set one or more reference lines or graphics for panel 3 matching, and set the error value;

[0049] S4: Create an alarm list, which includes communication alarms and field alarms, and set the alarm values ​​for the alarm status of each component, and set the number of alarms for conversion failure;

[0050] S5: Recognize the captured photo and retrieve the preset front view of the corresponding recognition area panel 3 for comparison with the photo. During the comparison, first match the position of panel 3 by reference point, and then compare the reference line or graphic.

[0051] S501: If the comparison results match, perform component alarm judgment and determine whether it is a transformation view. Otherwise, ignore it. If it is, judge the old and new transformation data. If the new transformation data is saved as the default transformation data, the old transformation data will not be changed.

[0052] S502: If the comparison results do not match, determine whether it is a view transformation. If it is, record the number of transformation failures and re-perform the view transformation. Otherwise, call the default transformation data for transformation.

[0053] S503: If there is no conversion data or the photo needs to be reconverted, retrieve the stereoscopic image of the corresponding panel 3, match the corresponding points in the photo and the stereoscopic image according to the reference point, and adjust the position of the stereoscopic image according to the reference line or graphic so that the view of panel 3 of the stereoscopic image matches and fits the captured photo. Then, obtain the conversion data of the photo according to the data of converting the stereoscopic image into the orthographic view, and retain the conversion data. The converted photo is then re-compared to the reference in S5.

[0054] S6: Compare the successfully captured photos with the alarm values, and issue alarms according to the alarm list for cases that reach the alarm value or the number of conversion failure alarms.

[0055] like Figure 1 and 2 As shown, this method can automatically identify panel areas and perform conversion recognition on panels at different angles. The conversion recognition uses 3D simulation conversion to eliminate the influence of viewing angle, increase the accuracy of automatic judgment of panel components, and can also automatically learn and adjust default conversion data to increase conversion efficiency. Conversion failures are recorded and alarms are issued based on the number of failures for timely adjustment. This enables 24-hour fast and accurate inspection, speeds up response time, eliminates the need to modify existing panels, is easy to install, has strong applicability, and can accurately trace the source of faults through photos, facilitating maintenance, eliminating manual inspection, reducing labor, and avoiding fatigue-related misjudgments. It can achieve identification and judgment at the minute or even second level.

[0056] like Figure 4 As shown, in one embodiment, S503 is followed by S504: setting the automatic overwrite time or number of photos to be saved, and timestamping and packaging the photos before and after conversion.

[0057] like Figure 3 As shown, in one embodiment, S0 is provided before S1: setting up ordinary accounts and administrator accounts. Ordinary accounts can access and view photos, as well as receive and control alarms, while administrator accounts have all permissions.

[0058] Example 2:

[0059] like Figure 5As shown, compared to Embodiment 1, the alarm value in S4 of this embodiment includes one or more combinations of displayed content, brightness, or color. In S6, false alarm and confirmed alarm functions are set. When an alarm occurs, the alarm type is marked as a false alarm or a confirmed alarm. If a false alarm is triggered within a time period or number of recognitions, the alarm value at the time of the false alarm is recorded as the false alarm value, and the false alarm value with the maximum difference from the original alarm value is automatically taken as the new alarm value. If no alarm is triggered within a time period or number of recognitions, the alarm value at the time of the correct alarm is recorded as the correct alarm value. The alarm value is automatically set to the midpoint between the correct value (with the smallest difference from the alarm value) and the alarm value itself. During use, the false alarm value with the largest difference is automatically replaced with the alarm value to avoid false alarms. After there are no false alarms, the alarm value is continuously and automatically corrected by the correct value with the smallest difference and the average of the alarm values, so that the alarm value is closer to the accurate trigger value. When the brightness and color are reduced due to the lifespan, it can be gradually corrected to an alarm value range that can be accurately judged, ensuring accurate alarm identification. The other methods and effects are the same as in Example 1.

[0060] In one embodiment, the S6 also includes a final endpoint function. The final endpoint is used to limit the alarm value endpoint corrected by the correct value. When the alarm value corrected by the correct value reaches the final endpoint, a fault alarm is triggered. By combining the final endpoint function with the function of continuously correcting the precise alarm value range, the device fault can be automatically determined when the corrected alarm value reaches the final endpoint, such as light not turning on, low brightness, and color change, which facilitates timely maintenance and replacement. The other methods and effects are the same as those in the aforementioned embodiments.

[0061] Example 3:

[0062] Compared to the previous embodiments, this embodiment includes an S001 before S1: A shooting device 1 and an AI monitoring station are set in the panel 3 area. The shooting device 1 is connected to the AI ​​monitoring station via wired or wireless means. The AI ​​monitoring station communicates with the mobile terminal via a PTZ. When in use, the mobile terminal can view and modify the AI ​​monitoring station in real time through the PTZ, which facilitates timely detection of problems. The remaining methods and effects are unchanged compared to the previous embodiments.

[0063] In one embodiment, before S1 there is S002: setting up an inspection track 4 according to the monitoring area of ​​panel 3, and mounting one or more shooting devices 1 on the inspection track 4 via track trolley 2;

[0064] S4 also sets the fixed stopping position and stopping time of the track trolley 2 at the designated panel 3;

[0065] In use, the camera device 1 can be set to move and inspect along the inspection track 4 with the track trolley 2, and stop at a fixed position to take pictures and identify. Reducing the number of camera devices 1 can increase the detection range of a single camera device 1, but will reduce the upper limit of the frequency of fixed-point shooting. The other methods and effects remain the same as those in the previous embodiments.

[0066] Example 4:

[0067] Compared to the previous embodiment, this embodiment has an additional S003 before S1: An angle adjustment track 5 parallel to the front of the panel 3 is set on the side of the inspection track 4 adjacent to the front of each panel 3. The angle adjustment track 5 is used for the movement of the track trolley 2 when the camera changes the camera angle. Multiple fixed stopping positions for taking pictures are set on the angle track.

[0068] S6 is followed by S7: When the alarm value is triggered, the track trolley 2 drives the shooting device 1 to take pictures from different front angles of the panel 3 for confirmation.

[0069] In use, by setting the parallel angle adjustment track 5, multi-angle shooting and recognition of the single panel 3 can be performed, which makes it easier to comprehensively judge whether the panel 3 is in a fault state or a false alarm state, further increasing the accuracy of automatic recognition. The other methods and effects are the same as those in the aforementioned embodiments.

[0070] Example 5:

[0071] Compared to the previous embodiments, this embodiment also includes an alarm self-checking function in S7. The self-checking function matches multi-angle recognition photos. When two or more photos have the same component that reaches the alarm value, an alarm is confirmed. If only one photo triggers the alarm value, it is marked as an incorrect shooting angle recognition photo. The erroneous photo is then marked, and the subsequent default shooting position is automatically changed to the remaining position on the angle adjustment track 5. In use, the self-checking function can reduce false alarms and intelligently change shooting positions that may cause false alarms, ensuring the accuracy of subsequent shooting recognition. The other methods and effects remain unchanged compared to the previous embodiments.

[0072] In one embodiment, after S7, there is also S701: the erroneous photo is sent to the administrator account, which has the function of correcting the erroneous photo to the correct recognition. When the erroneous photo is corrected to the correct recognition, photos taken from other angles are marked as having an incorrect shooting angle. The erroneous photo is marked and the subsequent shooting position is automatically changed back to the correct recognition position. In use, the administrator can manually correct the device's judgment error, avoiding the misjudgment of the self-verification function, so that the accuracy rate after learning and recognition is not less than 99.2%, the misjudgment rate is less than 0.7%, and the missed judgment rate is less than 0.1%. The other methods and effects are unchanged compared with the aforementioned embodiments.

[0073] In the description of this invention, it should be noted that the terms "upper", "lower", "left", "right", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0074] The control method of this invention is to control the device by manually starting and stopping the switch. The wiring diagram of the power element and the supply of power are common knowledge in the field. Since this invention is mainly used to protect mechanical devices, the control method and wiring layout will not be explained in detail.

[0075] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0076] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A control panel AI monitoring method, characterized in that, Includes the following steps: S1: Divide the panel (3) recognition area according to the monitoring screen of the shooting device (1), and upload the stereoscopic image and preset front view of the corresponding panel (3) according to the panel (3) type of the panel (3) recognition area. Then, label the panel (3) in the preset front view of each area with component areas. S2: Set the photo interval and number of shots; S3: Set the reference point for panel (3) position matching, and set more than one reference line or graphic for panel (3) matching, and set the error value; S4: Create an alarm list, which includes communication alarms and field alarms, and set the alarm values ​​for the alarm status of each component, and set the number of alarms for conversion failure; S5: Recognize the captured photo area and retrieve the preset front view of the corresponding recognition area panel (3) for comparison with the photo. When comparing, first match the position of the panel (3) by the reference point, and then compare the reference line or the graphic. S501: If the comparison results match, perform component alarm judgment and determine whether it is a transformation view. Otherwise, ignore it. If it is, judge the old and new transformation data. If the new transformation data is saved as the default transformation data, the old transformation data will not be changed. S502: If the comparison results do not match, determine whether it is a view transformation. If it is, record the number of transformation failures and re-perform the view transformation. Otherwise, call the default transformation data for transformation. S503: If there is no conversion data or the photo is re-converted, the stereoscopic image of the corresponding panel (3) is retrieved, the corresponding points in the photo and the stereoscopic image are matched according to the reference point, and the position of the stereoscopic image is adjusted according to the reference line or graphic so that the viewing angle of the panel (3) of the stereoscopic image matches the photo. Then, the conversion data of the photo is obtained according to the data of the stereoscopic image converted into the orthographic view, and the conversion data is retained. The converted photo is then compared with the reference in S5 again. S6: Compare the successfully captured photos with the alarm values, and issue alarms according to the alarm list for cases that reach the alarm value or the number of conversion failure alarms.

2. The AI ​​monitoring method for a control panel according to claim 1, characterized in that: S503 is followed by S504: setting the automatic overwrite time or number of photos to be saved, and timestamping and packaging the photos before and after conversion.

3. The AI ​​monitoring method for a control panel according to claim 2, characterized in that: Before S1, there is an S0: setting up ordinary accounts and administrator accounts. Ordinary accounts can access and view photos, as well as receive and control alarms, while administrator accounts have all permissions.

4. The AI ​​monitoring method for a control panel according to claim 1, characterized in that: The alarm value in S4 includes one or more combinations of displayed content, brightness, or color. In S6, false alarm and confirmed alarm functions are set. When an alarm occurs, the alarm type is marked as false alarm or confirmed alarm by the function. When a false alarm is triggered within a time period or number of recognitions, the alarm value at the time of the false alarm is recorded as the false alarm value, and the false alarm value with the maximum difference from the alarm value is automatically taken as the new alarm value. When no alarm is triggered within a time period or number of recognitions, the alarm value at the time of the correct alarm is recorded as the correct value, and the midpoint between the correct value with the minimum difference from the alarm value and the alarm value is automatically taken as the new alarm value.

5. The AI ​​monitoring method for a control panel according to claim 4, characterized in that: The S6 also includes a final endpoint function, which is used to limit the alarm value endpoint by correcting the correct value. When the alarm value corrected by the correct value reaches the final endpoint, a fault alarm is triggered.

6. The AI ​​monitoring method for a control panel according to claim 1, characterized in that: Before S1, there is S001: A shooting device (1) and an AI monitoring station are set in the panel (3) area. The shooting device (1) communicates with the AI ​​monitoring station via wired or wireless means. The AI ​​monitoring station communicates with the mobile terminal via a PTZ.

7. The AI ​​monitoring method for a control panel according to claim 6, characterized in that: Before S1, there is S002: according to the monitoring area of ​​the panel (3), the inspection track (4) is set up, and one or more shooting devices (1) are mounted on the inspection track (4) by the track trolley (2); The S4 also sets the fixed stopping position and stopping time of the track trolley (2) at the designated panel (3).

8. A control panel AI monitoring method according to claim 1 or 7, characterized in that: Before S1, there is S003: An angle adjustment track (5) parallel to the front of the panel (3) is set on the side of the inspection track (4) adjacent to the front of each panel (3). The angle adjustment track (5) is used for the movement of the track trolley (2) when the camera changes the camera angle. Multiple fixed stopping positions for taking pictures are set on the angle track. After S6, there is S7: When the alarm value is triggered, the track trolley (2) drives the shooting device (1) to take pictures from different front angles of the panel (3) for confirmation.

9. The AI ​​monitoring method for a control panel according to claim 8, characterized in that: The S7 also includes an alarm self-check function. The self-check function matches multi-angle recognition photos. When two or more photos have the same component and reach the alarm value, the alarm is confirmed. When only one photo triggers the alarm value, it is marked as an error in the shooting angle recognition. The photo is marked as the error and the subsequent default shooting position is automatically changed to the remaining position on the angle adjustment track (5).

10. The AI ​​monitoring method for a control panel according to claim 8, characterized in that: S7 is followed by S701: erroneous photos are sent to the administrator account. The administrator account has the function of correcting erroneous photos to be correctly identified. When the erroneous photos are corrected to be correctly identified, photos taken from other angles are marked as having incorrect shooting angles. This erroneous photo is marked and the subsequent shooting position is automatically changed back to the correct identification position.

Images