An intelligent drilling machine operation safety warning system integrating current detection and tool tracking functions

Abstract

The application provides an intelligent drilling machine operation safety early warning system integrating current detection and tool tracking functions, relates to the technical field of industrial automation safety monitoring, and camera collects real-time video streams of a drilling machine, a drill bit and a white glove, a current transformer collects current values of the drilling machine, an edge computing controller receives the processed real-time video streams, transmits them into a YOLO v5 target detection model for identification, if the edge computing controller detects that the drilling machine is in a working state and the drill bit and the white glove overlap in a two-dimensional space, the edge computing controller loads and runs a monocular depth running model to estimate the depth of the drill bit and the white glove in a three-dimensional space, and judges whether the drill bit and the white glove overlap in the three-dimensional space, and if overlapping, it is regarded as a violation.The application overcomes the limitations of single detection means in the prior art through current detection, target visual detection and monocular depth detection, improves the operation safety, and improves the intelligence and automation level of the early warning system.

Description

Technical Field

[0001] This invention relates to the field of industrial automation safety monitoring technology, specifically to an intelligent drilling machine operation safety early warning system that integrates current detection and tool tracking functions, which is particularly suitable for safe operation and risk control of drilling machines in industrial production environments. Background Technology

[0002] In existing technical solutions, the safety monitoring methods for drilling operations mainly rely on traditional mechanical protection measures and manual operation. Although some automated monitoring technologies have been introduced in recent years, these technologies still have many limitations in terms of safety, real-time performance, and intelligence.

[0003] In most industrial settings, drilling machine safety relies on mechanical safety devices such as protective covers, limit switches, and emergency braking systems. These devices act as physical barriers to prevent direct injury to operators from the drill bit during operation. However, the effectiveness of mechanical safety measures is limited. Their primary function is to prevent accidental contact and provide emergency stops in case of mechanical failure; they cannot proactively identify and respond to potentially hazardous behaviors during operation. The protective effect of mechanical safety devices is static and difficult to adapt to complex and dynamically changing working environments. Especially in high-intensity production operations, operators may unintentionally bypass or ignore these safety measures, leading to safety hazards. Furthermore, mechanical devices cannot achieve intelligent monitoring of the operation process, relying on the operator's experience and alertness. Over extended periods of operation, negligence can easily occur, leading to safety accidents.

[0004] Current detection technology is widely used in drilling machine operation monitoring. By monitoring changes in the current of the motor during operation, it indirectly determines whether the drill bit is in operation. This technology identifies changes in the drill machine's load through fluctuations in current data, thereby determining whether the drill machine is processing a workpiece. However, this approach only provides information on whether the drill bit is operating; it cannot provide comprehensive feedback on the drill bit's position, the surrounding environment, or other potential safety hazards during operation. Current detection technology only provides limited operational status information and cannot combine the actual position of the drill bit with the operator's behavior for a comprehensive judgment. Especially in situations involving multiple operators or complex workpieces, current fluctuations cannot accurately reflect the operational safety status. Furthermore, current detection can only monitor the mechanical state and cannot provide warnings about potential human actions (such as hand contact with dangerous areas), thus creating a significant safety blind spot.

[0005] In recent years, with the development of image processing and computer vision technologies, some monitoring systems have introduced camera-based visual recognition solutions, aiming to monitor drilling operations in real time through video streams. However, existing visual recognition systems are typically quite limited in function, mainly focusing on detecting whether the drill bit is within the work area or simply identifying whether operators have entered hazardous areas. The real-time performance and accuracy of these systems are easily affected by external environmental factors (such as lighting and occlusion), and they lack the ability to process three-dimensional depth information, thus failing to provide comprehensive operational safety assurance. Summary of the Invention

[0006] To address the shortcomings of the existing technologies, this invention provides an intelligent drilling machine operation safety early warning system that integrates current detection and tool tracking functions. This system enables comprehensive monitoring of tool status and operator behavior during drilling operations, overcoming the limitations of single detection methods in existing technologies. It improves operational safety while enhancing the intelligence and automation level of the monitoring system, making it particularly suitable for safe operation and risk control of drilling machines in industrial production environments.

[0007] This invention is achieved through the following technical solution:

[0008] A smart drilling machine operation safety early warning system integrating current detection and tool tracking functions is characterized by including a camera, an edge computing controller, a current transformer, etc. The camera is installed on the opposite column of the drilling machine, covering the entire drilling operation area.

[0009] The camera captures real-time video streams of the drilling machine, drill bit, and white gloves. During the acquisition process, the edge computing controller sends instructions to the camera via the ONVIF protocol to adjust the focus and viewing angle in real time to ensure that the drilling machine and drill bit are centered in the monitoring screen.

[0010] The current transformer is mounted on the live wire of the power supply box of the drill spindle motor to collect the drill current value and is connected to the power transmitter. The power transmitter is connected to the transparent transmission module, and the transparent transmission module is connected to the edge computing controller.

[0011] The edge computing controller receives real-time video streams and drill current values. After processing the real-time video streams, it inputs them into the YOLO v5 target vision detection model for identification of the drill, drill bit, and white glove. At the same time, the edge computing controller determines the working status of the drill through the drill current values. If the YOLO v5 target vision detection model identifies that the drill bit and white glove overlap in two-dimensional space and the drill is in working condition, the edge computing controller runs a monocular depth estimation model to detect whether the drill bit and white glove overlap in three-dimensional space. If they overlap, it is considered a violation.

[0012] Furthermore, the edge computing controller controls the camera to adjust its focus and viewing angle, specifically including: the edge computing controller controls the camera to move to preset point 1 for initial detection; the edge computing controller detects whether there is a drilling machine in the current monitoring screen; if there is no drilling machine, the edge computing controller controls the camera to move to the right until the drilling machine is detected in the monitoring screen; if the drilling machine is detected, the edge computing controller controls the camera to zoom in; since the drill bit is always located below the drilling machine, the edge computing controller controls the camera to move downwards until the drill bit is found; after finding the drill bit, the edge computing controller calculates the offset of the drill bit in the monitoring screen and controls the camera to adjust according to the offset so that the drill bit is located in the center of the monitoring screen.

[0013] Furthermore, the offset calculation formula is as follows:

[0014]

[0015] Where x1, y1 are the coordinates of the drill bit at the top left corner of the monitoring screen, x2, y2 are the coordinates of the drill bit at the bottom right corner of the monitoring screen, boxCenterX is the X-axis coordinate of the center point of the drill bit in the monitoring screen, boxCenterY is the Y-axis coordinate of the center point of the drill bit in the monitoring screen, imgShapeX is the total width of the monitoring screen on the X-axis, which is also the pixel width of the image, imgShapeY is the total height of the monitoring screen on the Y-axis, which is also the pixel height of the image, offsetX is the offset of the center point of the drill bit relative to the center point of the monitoring screen on the X-axis, and offsetY is the offset of the center point of the drill bit relative to the center point of the monitoring screen on the Y-axis.

[0016] Furthermore, the calculation of the overlap between the drill bit and the white glove in two-dimensional space is as follows:

[0017] The YOLO v5 object vision detection model generates a bounding box for each detected object. The bounding box is represented by four parameters: center point coordinates (x, y), width w, and height h.

[0018] First, we need to calculate the intersection area of ​​the drill bit in bounding box A and the white glove in bounding box B. If the coordinates of the drill bit are (x1, y1, x2, y2) and the coordinates of the white glove are (x1', y1', x2', y2'), the top-left corner of the intersection is (max(x1, x1'), max(y1, y1')), and the bottom-right corner is (min(x2, x2'), min(y2, y2')). The calculation formula is as follows:

[0019] Intersection Area=max(0,min(x2,x2′)-max(x1,x1′))*max(0,min(y2-y2′)-max(y1,y1′));

[0020] Next, calculate the area of ​​the union of the drill bit in bounding box A and the white glove in bounding box B. The calculation formula is as follows:

[0021] Union Area=(x2-x1)*(y2-y1)+(x2′-x1′)*(y2′-t1′)-Intersection Area;

[0022] Finally, calculate the intersection-union ratio (CUC) of the overlap between the drill bit in bounding box A and the white glove in bounding box B. The calculation formula is as follows:

[0023]

[0024] Set an IoU threshold. If the IoU is greater than the threshold, it is determined that they overlap in two-dimensional space.

[0025] Furthermore, the IoU threshold is set to 0.2.

[0026] Furthermore, the calculation of the overlap between the drill bit and the white glove in three-dimensional space is as follows:

[0027] When the drill bit and the white glove overlap in two-dimensional space, the edge computing controller takes a screenshot and sends the screenshot to the monocular depth estimation model for calculation. The depth estimation model calculates the quantized depth value qnt. First, the quantized depth value is dequantized and converted into a floating-point depth value. The calculation formula is as follows:

[0028] DeQntValue=(qnt-zp)*scale;

[0029] Where DeQnValue is the dequantized floating-point depth value, zp is the zero offset used to adjust the quantized baseline; scale is the quantization scaling factor used to scale the quantized value back to a floating-point number.

[0030] Next, the floating-point depth values ​​converted from dequantization are normalized to obtain the depth map. The calculation formula is as follows:

[0031]

[0032] In depth map processing, DepthMap[h][w] represents the depth value of the pixel located in row h and column w in the depth map, DepthMin and DepthMax are the minimum and maximum depth values ​​in the depth map, and MaxValue is the normalized maximum value, set to 255;

[0033] The center coordinates of the drill bit and the white glove are calculated using the following formula:

[0034]

[0035] top_left and bottom_right are the coordinates of the top left and bottom right corners of the coordinate box in the screenshot, respectively.

[0036] The center coordinates of the drill bit (centerA(x,y)) and the white glove (centerB(x,y)) are calculated. Then, they are indexed in the DepthMap[h][w] to obtain the depth values ​​of the center coordinates of the drill bit and the glove. Finally, the center depth value is calculated using the following formula:

[0037] Depth Difference = abs(D_drill bit - D_white glove);

[0038] Where DepthDifference is the center depth value, abs represents the absolute value, D_drill_bit is the depth value of the center coordinates of the drill bit, and D_white_glove is the depth value of the center coordinates of the white_glove.

[0039] Set a center depth threshold. If the absolute value of the center depth is less than the center depth threshold, it is determined that the drill bit and the white glove have a three-dimensional overlap.

[0040] Furthermore, the center depth threshold is set to 2.5.

[0041] Furthermore, the camera transmits the real-time video stream to the edge computing controller via the RTSP protocol, and the edge computing controller calls the hardware-accelerated H.264 decoder for real-time decoding.

[0042] Furthermore, if the detected drilling machine current value is greater than 0.6A, the drilling machine is determined to be in working condition; if the detected rotating bed current value is less than 0.6A, the drilling machine is determined to be in stationary condition.

[0043] Furthermore, the system also includes an audible and visual alarm and a remote monitoring platform. The edge computing controller is connected to the audible and visual alarm and the remote monitoring platform respectively. If it is determined that the drill bit and the white glove overlap in three-dimensional space, it is considered a violation. The edge computing controller then triggers the audible and visual alarm and uploads the violation image to the remote monitoring platform in real time.

[0044] The working principle of this invention is as follows:

[0045] The camera captures real-time video streams of the drilling machine, drill bit, and white gloves. During the acquisition process, the edge computing controller sends instructions to the camera via the ONVIF protocol to adjust the focus and viewing angle in real time, ensuring that the drilling machine and drill bit are centered in the monitoring screen, and transmits the real-time video stream to the edge computing controller.

[0046] The current transformer collects the drilling machine current value and connects to the power transmitter. The power transmitter is connected to the transparent transmission module, which in turn is connected to the edge computing controller. The drilling machine current value is transmitted to the edge computing controller to determine whether the drilling machine is in working condition.

[0047] After receiving and processing the real-time video stream, the edge computing controller inputs the YOLO v5 object detection model to identify the drilling machine, drill bit, and white glove. If the edge computing controller detects that the drilling machine is in operation and the drill bit and white glove overlap in two-dimensional space, the edge computing controller loads and runs a monocular depth estimation model to estimate the depth of the drill bit and white glove in three-dimensional space, thereby determining whether the drill bit and white glove overlap in three-dimensional space. If they overlap, it is considered a violation.

[0048] The beneficial effects of this invention are as follows:

[0049] 1. This invention introduces a multimodal fusion technology of current detection, target visual detection, and monocular depth detection to form a high-precision early warning system. This system can more comprehensively perceive the working environment, overcome the limitations of single detection methods in existing technologies, improve operational safety, and enhance the intelligence and automation level of the early warning system.

[0050] 2. This invention employs a current transformer and LoRa communication technology to avoid direct connection to the circuit, thus not affecting the original circuit, reducing installation difficulty and safety risks. Furthermore, the current transformer has good linearity and wide range characteristics, enabling it to capture minute changes in current and improve detection sensitivity. The LoRa communication technology enables long-distance, low-power, and highly reliable real-time data transmission.

[0051] 3. This invention can effectively determine the start-up, stop, or overload status of the drilling machine through real-time current detection. Combined with visual recognition results, current detection provides a reliable basis for determining whether the drill bit is in working condition, thereby improving the system's judgment capability and avoiding safety hazards caused by visual misjudgment.

[0052] 4. This invention achieves intelligent control of the camera through the ONVIF protocol, enabling it to dynamically adjust the viewing angle and focus, always keeping the drilling machine and drill bit within the optimal monitoring range. This not only improves the flexibility of monitoring but also ensures continuous monitoring of important workstations, reducing safety hazards caused by blind spots in monitoring.

[0053] 5. This invention, through the linkage mechanism of the audible and visual alarm and the remote monitoring platform, can not only trigger alarms on-site, but also upload images of violations to the remote monitoring platform in real time, facilitating remote intervention and handling by management personnel. This dual monitoring mechanism effectively improves the safety of the operation process and the controllability of the system. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating the workflow of the early warning system in an embodiment of the present invention. Detailed Implementation

[0055] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1

[0057] A smart drilling machine operation safety early warning system integrating current detection and tool tracking functions includes a camera, an edge computing controller, a current transformer, etc. The camera is installed on the column opposite the drilling machine, covering the entire drilling operation area.

[0058] The camera captures real-time video streams of the drilling machine, drill bit, and white gloves. During the acquisition process, the edge computing controller sends commands to the camera via the ONVIF protocol to adjust the focus and viewing angle in real time, ensuring that the drilling machine and drill bit are centered in the monitoring screen.

[0059] The current transformer is mounted on the live wire of the power supply box of the drill spindle motor to collect the drill current value and is connected to the power transmitter. The power transmitter is connected to the transparent transmission module, and the transparent transmission module is connected to the edge computing controller.

[0060] In this implementation, the power transmitter converts the analog quantity collected by the current transformer into a digital quantity. The pass-through module is first connected to the power transmitter through the 485 interface to pass through the current value collected by the current transformer using LORA. The pass-through module is then connected to the edge computing controller through the 485 interface, thereby enabling the edge computing controller to non-intrusively collect the drilling machine current value and thus determine the working status of the drilling machine.

[0061] like Figure 1 As shown, the edge computing controller receives real-time video streams and drill current values. After processing the real-time video streams, it inputs them into the YOLO v5 target vision detection model for identification of the drill, drill bit, and white glove. At the same time, the edge computing controller determines the working status of the drill by the drill current value. If the YOLO v5 target vision detection model identifies that the drill bit and white glove overlap in two-dimensional space and the drill is in working condition, the edge computing controller runs a monocular depth estimation model to detect whether the drill bit and white glove overlap in three-dimensional space. If they overlap, it is considered a violation.

[0062] Furthermore, if the detected drilling machine current value is greater than 0.6A, the drilling machine is determined to be in working condition; if the detected rotating bed current value is less than 0.6A, the drilling machine is determined to be in stationary condition.

[0063] In this embodiment, the camera transmits the real-time video stream to the edge computing controller via the RTSP protocol, and the edge computing controller calls the hardware-accelerated H.264 decoder for real-time decoding.

[0064] Example 2

[0065] This embodiment further elaborates and supplements the implementation of the present invention based on Embodiment 1.

[0066] The edge computing controller controls the camera to adjust its focus and viewing angle, specifically by: the edge computing controller moving the camera to a preset point 1 for initial detection; the edge computing controller detecting whether a drilling machine exists in the current monitoring frame; if no drilling machine exists, controlling the camera to move to the right until the drilling machine is detected in the monitoring frame; if the drilling machine is detected, the edge computing controller controlling the camera to zoom in; since the drill bit is always below the drilling machine, controlling the camera to move downwards until the drill bit is found; after finding the drill bit, the edge computing controller calculating the offset of the drill bit in the monitoring frame and controlling the camera to adjust according to the offset so that the drill bit is in the center of the monitoring frame.

[0067] The offset calculation formula is as follows:

[0068]

[0069] Where x1, y1 are the coordinates of the drill bit at the top left corner of the monitoring screen, x2, y2 are the coordinates of the drill bit at the bottom right corner of the monitoring screen, boxCenterX is the X-axis coordinate of the center point of the drill bit in the monitoring screen, boxCenterY is the Y-axis coordinate of the center point of the drill bit in the monitoring screen, imgShapeX is the total width of the monitoring screen on the X-axis, which is also the pixel width of the image, imgShapeY is the total height of the monitoring screen on the Y-axis, which is also the pixel height of the image, offsetX is the offset of the center point of the drill bit relative to the center point of the monitoring screen on the X-axis, and offsetY is the offset of the center point of the drill bit relative to the center point of the monitoring screen on the Y-axis.

[0070] The process of calculating the overlap between the drill bit and the white glove in two-dimensional space is as follows:

[0071] The YOLO v5 object vision detection model generates a bounding box for each detected object. The bounding box is represented by four parameters: center point coordinates (x, y), width w, and height h.

[0072] First, we need to calculate the intersection area of ​​the drill bit in bounding box A and the white glove in bounding box B. If the coordinates of the drill bit are (x1, y1, x2, y2) and the coordinates of the white glove are (x1', y1', x2', y2'), the top-left corner of the intersection is (max(x1, x1'), max(y1, y1')), and the bottom-right corner is (min(x2, x2'), min(y2, y2')). The calculation formula is as follows:

[0073] Intersection Area=max(0,min(x2,x2′)-max(x1,x1′))*max(0,min(y2-y2′)-max(y1,t1′));

[0074] Next, calculate the area of ​​the union of the drill bit in bounding box A and the white glove in bounding box B. The calculation formula is as follows:

[0075] Union Area=(x2-x1)*(y2-y1)+(x2′-x1′)*(y2′-y1′)-Intersection Area;

[0076] Finally, calculate the intersection-union ratio (CUC) of the overlap between the drill bit in bounding box A and the white glove in bounding box B. The calculation formula is as follows:

[0077]

[0078] Set an IoU threshold. If the IoU is greater than the threshold, it is determined that they overlap in two-dimensional space.

[0079] Furthermore, the IoU threshold is set to 0.2.

[0080] Furthermore, the calculation of the overlap between the drill bit and the white glove in three-dimensional space is as follows:

[0081] When the drill bit and the white glove overlap in two-dimensional space, the edge computing controller takes a screenshot and sends the screenshot to the monocular depth estimation model for calculation. The depth estimation model calculates the quantized depth value qnt. First, the quantized depth value is dequantized and converted into a floating-point depth value. The calculation formula is as follows:

[0082] DeQntValue=(qnt-zp)*scale;

[0083] Where DeQnValue is the dequantized floating-point depth value, zp is the zero offset used to adjust the quantized baseline; scale is the quantization scaling factor used to scale the quantized value back to a floating-point number.

[0084] Next, the floating-point depth values ​​converted from dequantization are normalized to obtain the depth map. The calculation formula is as follows:

[0085]

[0086] In depth map processing, DepthMap[h][w] represents the depth value of the pixel located in row h and column w in the depth map, DepthMin and DepthMax are the minimum and maximum depth values ​​in the depth map, and MaxValue is the normalized maximum value, set to 255;

[0087] The center coordinates of the drill bit and the white glove are calculated using the following formula:

[0088]

[0089] top_left and bottom_right are the coordinates of the top left and bottom right corners of the coordinate box in the screenshot, respectively.

[0090] The center coordinates of the drill bit (centerA(x,y)) and the white glove (centerB(x,y)) are calculated. Then, they are indexed in the DepthMap[h][w] to obtain the depth values ​​of the center coordinates of the drill bit and the glove. Finally, the center depth value is calculated using the following formula:

[0091] Depth Difference = abs(D_drill bit - D_white glove);

[0092] Where DepthDifference is the center depth value, abs represents the absolute value, D_drill_bit is the depth value of the center coordinates of the drill bit, and D_white_glove is the depth value of the center coordinates of the white_glove.

[0093] Set a center depth threshold. If the absolute value of the center depth is less than the center depth threshold, it is determined that the drill bit and the white glove have a three-dimensional overlap.

[0094] Furthermore, the center depth threshold is set to 2.5.

[0095] Example 3

[0096] This embodiment further elaborates and supplements the implementation of the present invention based on Embodiment 1 or Embodiment 2.

[0097] The system also includes an audible and visual alarm and a remote monitoring platform. An edge computing controller is connected to both the alarm and the platform. If the drill bit and white gloves are determined to overlap in three-dimensional space, this is considered a violation. The edge computing controller then triggers the audible and visual alarm and uploads the violation image to the remote monitoring platform in real time. This linkage mechanism between the alarm and the remote monitoring platform not only triggers alarms on-site but also uploads alarm images to the platform in real time, facilitating remote intervention and handling by management personnel and improving the safety and controllability of the operation.

Claims

1. An intelligent drilling machine operation safety warning system integrating current detection and tool tracking functions, characterized in that: Includes cameras, edge computing controllers, current transformers, etc. The camera is installed on the column opposite the drilling machine, covering the entire area where the drilling machine is operating; The camera captures real-time video streams of the drilling machine, drill bit, and white gloves. During the acquisition process, the edge computing controller sends instructions to the camera via the ONVIF protocol to adjust the focus and viewing angle in real time to ensure that the drilling machine and drill bit are centered in the monitoring screen. The current transformer is mounted on the live wire of the power supply box of the drill spindle motor to collect the drill current value and is connected to the power transmitter. The power transmitter is connected to the transparent transmission module, and the transparent transmission module is connected to the edge computing controller. The edge computing controller receives real-time video streams and drill current values. After processing the real-time video streams, it inputs them into the YOLO v5 target vision detection model for identification of the drill, drill bit, and white glove. At the same time, the edge computing controller determines the working status of the drill through the drill current values. If the YOLO v5 target vision detection model identifies that the drill bit and white glove overlap in two-dimensional space and the drill is in working condition, the edge computing controller runs a monocular depth estimation model to detect whether the drill bit and white glove overlap in three-dimensional space. If they overlap, it is considered a violation.

2. The intelligent drilling machine operation safety warning system integrated with current detection and tool tracking functions according to claim 1, characterized in that: The edge computing controller controls the camera to adjust its focus and viewing angle, specifically by: the edge computing controller moving the camera to a preset point 1 for initial detection; the edge computing controller detecting whether a drilling machine exists in the current monitoring frame; if no drilling machine exists, controlling the camera to move to the right until the drilling machine is detected in the monitoring frame; if the drilling machine is detected, the edge computing controller controlling the camera to zoom in; since the drill bit is always below the drilling machine, controlling the camera to move downwards until the drill bit is found; after finding the drill bit, the edge computing controller calculating the offset of the drill bit in the monitoring frame and controlling the camera to adjust according to the offset so that the drill bit is in the center of the monitoring frame.

3. The intelligent drilling machine operation safety early warning system integrating current detection and tool tracking functions as described in claim 2, characterized in that: The offset calculation formula is as follows: Where x1, y1 are the coordinates of the drill bit at the top left corner of the monitoring screen, x2, y2 are the coordinates of the drill bit at the bottom right corner of the monitoring screen, boxCenterX is the X-axis coordinate of the center point of the drill bit in the monitoring screen, boxCenterY is the Y-axis coordinate of the center point of the drill bit in the monitoring screen, imgShapeX is the total width of the monitoring screen on the X-axis, which is also the pixel width of the image, imgShapeY is the total height of the monitoring screen on the Y-axis, which is also the pixel height of the image, offsetX is the offset of the center point of the drill bit relative to the center point of the monitoring screen on the X-axis, and offsetY is the offset of the center point of the drill bit relative to the center point of the monitoring screen on the Y-axis.

4. The intelligent drilling machine operation safety warning system integrated with current detection and tool tracking functions as claimed in claim 1, characterized in that: The process of calculating the overlap between the drill bit and the white glove in two-dimensional space is as follows: The YOLO v5 object vision detection model generates a bounding box for each detected object. The bounding box is represented by four parameters: center point coordinates (x, y), width w, and height h. First, we need to calculate the intersection area of ​​the drill bit in bounding box A and the white glove in bounding box B. If the coordinates of the drill bit are (x1, y1, x2, y2) and the coordinates of the white glove are (x1', y1', x2', y2'), the top-left corner of the intersection is (max(x1, x1'), max(y1, y1')), and the bottom-right corner is (min(x2, x2'), min(y2, y2')). The calculation formula is as follows: Intersection Area=max(0,min(x2,x2′)-max(x1,x1′))*max(0,min(y2-y2′)-max(y1,y1′)); Next, calculate the area of ​​the union of the drill bit in bounding box A and the white glove in bounding box B. The calculation formula is as follows: Union Area=(x2-x1)*(y2-y1)+(x2'-x1')*(y2'-y1)-Intersection Area; Finally, calculate the intersection-union ratio (CUC) of the overlap between the drill bit in bounding box A and the white glove in bounding box B. The calculation formula is as follows: Set an IoU threshold. If the IoU is greater than the threshold, it is determined that they overlap in two-dimensional space.

5. The intelligent drilling machine operation safety warning system integrated with current detection and tool tracking functions as claimed in claim 4, characterized in that: The IoU threshold is set to 0.

2.

6. The intelligent drilling machine operation safety warning system integrated with current detection and tool tracking functions as claimed in claim 1, wherein: The calculation process of the drill bit and white glove overlapping in three-dimensional space is as follows: When the drill bit and the white glove overlap in two-dimensional space, the edge computing controller takes a screenshot and sends the screenshot to the monocular depth estimation model for calculation. The depth estimation model calculates the quantized depth value qnt. First, the quantized depth value is dequantized and converted into a floating-point depth value. The calculation formula is as follows: DeQntValue=(qnt-zp)*scale; Where DeQnValue is the dequantized floating-point depth value, zp is the zero offset used to adjust the quantized baseline; scale is the quantization scaling factor used to scale the quantized value back to a floating-point number. Next, the floating-point depth values ​​converted from dequantization are normalized to obtain the depth map. The calculation formula is as follows: In depth map processing, DepthMap[h][w] represents the depth value of the pixel located in row h and column w in the depth map, DepthMin and DepthMax are the minimum and maximum depth values ​​in the depth map, and MaxValue is the normalized maximum value, set to 255; The center coordinates of the drill bit and the white glove are calculated using the following formula: top_left and bottom_right are the coordinates of the top left and bottom right corners of the coordinate box in the screenshot, respectively. The center coordinates of the drill bit (centerA(x,y)) and the white glove (centerB(x,y)) are calculated. Then, they are indexed in the DepthMap[h][w] to obtain the depth values ​​of the center coordinates of the drill bit and the glove. Finally, the center depth value is calculated using the following formula: Depth Difference = abs(D_drill bit - D_white glove); Where DepthDifference is the center depth value, abs represents the absolute value, D_drill_bit is the depth value of the center coordinates of the drill bit, and D_white_glove is the depth value of the center coordinates of the white_glove. Set a center depth threshold. If the absolute value of the center depth is less than the center depth threshold, it is determined that the drill bit and the white glove have a three-dimensional overlap.

7. The intelligent drilling machine operation safety warning system integrated with current detection and tool tracking functions as claimed in claim 6, wherein: The center depth threshold is set to 2.

5.

8. The intelligent drilling machine operation safety warning system integrated with current detection and tool tracking functions as claimed in claim 1, wherein: The camera transmits real-time video streams to the edge computing controller via the RTSP protocol, and the edge computing controller calls a hardware-accelerated H.264 decoder for real-time decoding.

9. The integrated current detection and tool tracking function's smart drilling machine operation safety warning system according to claim 1, characterized in that: If the detected drilling machine current value is greater than 0.6A, the drilling machine is determined to be in working condition. If the detected rotary table current value is less than 0.6A, the drilling machine is considered to be in a stationary state.

10. The intelligent drilling machine operation safety warning system integrated with current detection and tool tracking functions as claimed in claim 1, wherein: The system also includes an audible and visual alarm and a remote monitoring platform. The edge computing controller is connected to the audible and visual alarm and the remote monitoring platform respectively. If it is determined that the drill bit and the white glove overlap in three-dimensional space, it is considered a violation. The edge computing controller will then trigger the audible and visual alarm and upload the violation image to the remote monitoring platform in real time.

Images