초록 ▼

An infrared vision sensing detection method and device for narrow-gap weld seam deviation are provided. The device includes a shaking (or rotating) arc narrow-gap welding torch, an arc current sensor, a computer image processing system, and an infrared photographing system. The infrared photographin

An infrared vision sensing detection method and device for narrow-gap weld seam deviation are provided. The device includes a shaking (or rotating) arc narrow-gap welding torch, an arc current sensor, a computer image processing system, and an infrared photographing system. The infrared photographing system includes an infrared camera which acquires an infrared image of a welding region in an external triggering manner when an arc shakes (or rotates) to a position closest to the left side wall or right side wall of a groove. After computer image processing, a welding wire position and a groove edge information is extracted in real time, and a weld seam deviation is calculated according to position changes of a welding wire relative to the left side wall and the right side wall of the groove, and the weld seam deviation is output. During pulsed arc welding, a signal in a base value period of the arc current pulse is detected by using the current sensor, thereby realizing welding image acquisition synchronized with the base value current period of the pulsed arc.

대표청구항 ▼

1. An infrared vision sensing detection device for narrow-gap weld seam deviation, comprising a narrow-gap welding torch, an arc current sensor, a computer image processing system, and an infrared photographing system, wherein one end of a bent conducting rod in the narrow-gap welding torch is conne

1. An infrared vision sensing detection device for narrow-gap weld seam deviation, comprising a narrow-gap welding torch, an arc current sensor, a computer image processing system, and an infrared photographing system, wherein one end of a bent conducting rod in the narrow-gap welding torch is connected to a motor driving and a feeding mechanism, and the other end is connected to a straight-type contact tube, a welding wire passing through the narrow-gap welding torch extends into a groove to be welded to generate a welding arc, and the infrared photographing system comprises a digital infrared camera and an infrared filtering system, wherein: the computer image processing system comprises an image acquisition card, a welding wire position information extraction module and a weld seam deviation value calculation module that are sequentially connected, and the image acquisition card is connected to the infrared camera through a video line, a signal PL or P1 is input to the infrared camera and the image acquisition card simultaneously when the arc generated by the motor driving and the feeding mechanism moves to a position closest to the left or right side wall of the groove, one end of the welding power source is connected to the motor driving and the feeding mechanism, a connection cable on the other end passes through a detection ring of the arc current sensor or is connected in series with the arc current sensor, and then is connected to a workpiece, a pulsed welding arc base value current signal ib detected by the arc current sensor and the arc position signal PL or PR jointly act on an image photographing trigger signal input end of the infrared camera. 2. A detection method of the infrared vision sensing detection device for narrow-gap weld seam deviation according to claim 1, comprising the following steps: 1) the entire narrow-gap welding torch and the infrared photographing system moving towards the front of the groove at a welding speed Vw together, triggering, according to the arc position signal PL or PR, the infrared camera to acquire a welding region image at this time point and send the welding region image to the computer image processing system through the image acquisition card;2) when the arc moves to a position closest to the left or right side wall of the groove, the welding wire position information extraction module receiving a welding image information from the image acquisition card, extracting, by processing the welding region image, a current distance X1i from a right position detection point of the welding wire to a groove left edge and a current distance X2i from a left position detection point of the welding wire to a groove right edge, and sending the current distances to the weld seam deviation value calculation module; and3) the weld seam deviation value calculation module calculating a current detection value ΔXi of the weld seam deviation according to the formula ΔXi=(X1i−X2i)/2, and then using a median or a mean of the latest n (n≧1) detection values of the weld seam deviation as a current sampling value ΔXsi of the weld seam deviation, wherein if ΔXsi=0, then the weld seam is not deviated, if ΔXsi>0, then the welding wire is deviated towards the right side of the groove, and if ΔXsi<0, then the welding wire is deviated towards the left side of the groove. 3. The detection method according to claim 2, wherein in step 2), when the arc moves to the position closest to the right side wall of the groove, the welding wire position information extraction module adjusts an abscissa value of a positioning point B1i of a groove left image capture window according to a horizontal position change of a groove left edge line, to capture a groove left image without interference from the arc; after the groove left edge is extracted, a current distance L2i from the groove left edge to a left boundary of a global image is calculated; after a welding wire right position image is captured in a welding wire right image capture window, a current distance L1i from a center of the welding wire to the left boundary of the global image is extracted, to calculate a current distance from the right position detection point of the welding wire to the groove left edge: X1i=(L1i−L2i); when the arc moves to a position closest to a left side wall of the groove, the welding wire position information extraction module adjusts an abscissa value of a positioning point B2i of a groove right image capture window according to a horizontal position change of a groove right edge line, to capture a groove right image without interference from the arc; after the groove right edge is extracted, a current distance L3i from the groove right edge to the left boundary of the global image is calculated; after a welding wire left position image is captured in a welding wire left image capture window, a current distance L4i from the center of the welding wire to the left boundary of the global image is extracted, to calculate a current distance from the left position detection point of the welding wire to the groove right edge: X2i=(L3i−L4i). 4. The detection method according to claim 3, wherein the current distance L1i or L4i from the center of the welding wire to the left boundary of the global image is extracted first, and then the current distance L2i or L3i from the groove left or right edge to the left boundary of the global image is calculated. 5. The detection method according to claim 2, wherein when the arc moves to the position closest to the right side wall of the groove again, a next distance L2(i+1) from the groove left edge to the left boundary of the global image and a next distance L1(i+1) from the right position detection point of the welding wire to the left boundary of the global image are extracted first, and then a next distance from the right position detection point of the welding wire to the groove left edge is calculated: X1(i+1)=(L1(i+1)−L2(i+1)), thereby calculating a next detection value of the weld seam deviation: ΔX(i+1)=(X1(i+1)−X2i)/2, and so forth, so that the weld seam deviation is detected twice in one arc movement cycle. 6. The detection method according to claim 3, wherein ordinate values of the L2i and L2(i+1) detection points on the groove left edge line are the same as an ordinate value of the L3i detection point on the groove right edge line, ordinate values of the L1i and L1(i+1) detection points on the right-position welding wire axis are the same as an ordinate value of the L4i detection point on the left-position welding wire axis, a final detection value of the distance L1i, L4i or L1(i+1) from the welding wire position detection point to the left boundary of the global image is a median or a mean of detection values of m (m≧1) different positions that are in the welding wire image left and right capture windows and on the axis of the welding wire, and a final detection value of the distance L2i, L3i or L2(i+1) from the groove left and right edges to the left boundary of the global image is a median or a mean of detection values at k (k≧1) different positions that are in the groove left and right image capture windows and on the groove left and right edge lines. 7. The detection method according to claim 2, wherein: when the welding region image is processed, coordinate values of a highest point in an arc region are first extracted by means of global welding image processing, and coordinate values of positioning points of the welding wire left and right image capture windows are adjusted according to a change in the coordinate position of the highest point of the arc region;a small-window groove image captured by the groove left and right image capture windows and a small-window welding wire image captured by the welding wire left and right image capture windows are separately processed;during processing of the small-window images captured by the welding wire left and right image capture windows, a local adaptive threshold segmentation processing is performed first, and after the contour of the welding wire is extracted by means of morphological erosion on a full-window image, the framework of the welding wire is extracted by using a Canny edge detection algorithm, and finally, the axis position of the welding wire is calculated. 8. The detection method according to claim 3, wherein in the case of shaking or rotating pulsed arc welding, when the arc moves to the position closest to the left side wall or a right side wall of the groove, once the current sensor detects that the first pulsed welding arc base value current signal ib of the pulsed arc arrives, the infrared camera is triggered immediately, so as to acquire a welding region image with smallest interference from arc light at this time point, thereby achieving welding image acquisition synchronized with a base value current period of the pulsed arc. 9. The detection method according to claim 3, wherein when the arc moves to the position closest to the right side wall of the groove again, a next distance L2(i+1) from the groove left edge to the left boundary of the global image and a next distance L1(i+1) from the right position detection point of the welding wire to the left boundary of the global image are extracted first, and then a next distance from the right position detection point of the welding wire to the groove left edge is calculated: X1(i+1)=(L1(i+1)−L2(i+1)), thereby calculating a next detection value of the weld seam deviation: ΔX(i+1)=(X1(i+1)−X2i)/2, and so forth, so that the weld seam deviation is detected twice in one arc movement cycle. 10. The detection method according to claim 8, wherein ordinate values of the L2i and L2(i+1) detection points on the groove left edge line are the same as an ordinate value of the L3i detection point on the groove right edge line, ordinate values of the L1i and L1(i+1) detection points on the right-position welding wire axis are the same as an ordinate value of the L4i detection point on the left-position welding wire axis, a final detection value of the distance L1i, L4i or L1(i+1) from the welding wire position detection point to the left boundary of the global image is a median or a mean of detection values of m (m≧1) different positions that are in the welding wire image left and right capture windows and on the axis of the welding wire, and a final detection value of the distance L2i, L3i or L2(i+1) from the groove left and right edges to the left boundary of the global image is a median or a mean of detection values at k (k≧1) different positions that are in the groove left and right image capture windows and on the groove left and right edge lines. 11. The detection method according to claim 3, wherein: when the welding region image is processed, coordinate values of a highest point in an arc region are first extracted by means of global welding image processing, and coordinate values of positioning points of the welding wire left and right image capture windows are adjusted according to a change in the coordinate position of the highest point of the arc region;a small-window groove image captured by the groove left and right image capture windows and a small-window welding wire image captured by the welding wire left and right image capture windows are separately processed;during processing of the small-window images captured by the welding wire left and right image capture windows, a local adaptive threshold segmentation processing is performed first, and after the contour of the welding wire is extracted by means of morphological erosion on a full-window image, the framework of the welding wire is extracted by using a Canny edge detection algorithm, and finally, the axis position of the welding wire is calculated.