Research On Visual Recognition Technology Of Die Laser Repair

Moulds are important equipment and have been widely applied in plastic products and metal products in modern industry. Owing to the high efficiency, superior product quality and lower cost of production, moulds have occupied an extremely important position in the development of modern manufacturing. The overall level of manufacturing moulds in our country falls far behind the advanced countries due to low accuracy and low service life. Because of thermal failure, which is one of the important failure performance, hot working dies waste RMB of several billion yuan. Therefore, repairing smaller thermal cracks by laser melting and larger thermal cracks by laser filler wire have import academic values and perfect economic benefits. Coaxial visual detection helps to strengthen capability of laser moving, but does nothing to recognize weld pool depth,wire space position and other 3D information. A paraxial vision sensing system is added to detect weld pool width, wire location and defocusing amount to monitor repair quality.Double CCD vision systems was established to acquire relevant feature of laser mold repair processing. First of all, the model of radiative image and calibration was established to determine the mutual position relationship of imaging plane CCD and object. Crack features such as the roundness of the welt pool and gray feature of ROI molten are extracted by coaxial visual; the angle between welding wire and mould surface and the defocusing amount are extracted by the paraxial visual. NC code which generated by UG NX based on the crack features worked well in the processing of laser repairing. Actual dies was repaired by using the CCD vision sensor system to control the parameters of ROI regional feature, defocusing amount and wire angle, the manufacturing verification test demonstrated the effectiveness for improving the die service life.