The Mapping Model Research On Flexible Roll Forming Based On Visual Feedback

Flexible roll-forming, which takes the steel sheet or coil as the raw material by passing through multiple passes assembly with certain shape and variations according to some rules gradually to manufacture variable cross-section product, is a new plastic forming method of shaft parts. Compared with traditional constant product, flexible cross-section product have the advantages of strong load-bearing capacity, light weight structure, lower production cost and so on, which can better meet the actual production requirements and have promising application prospects in automotive, furniture and other fields.In the flexible roll forming process, the complexity of the movement control and the high required precision of rolls position and velocity make higher request for the servo motor control system to achieve smooth processing. In the forming process, there are many facts effecting forming precision. Plate is subject to lateral bending, tensile deformation, longitudinal tensile and vertical shear deformation, and each passes of forming force is a real-time changes, so in order to ensure that movement trajectory of each passes roll can match the outline of the plate in time and space, the forming process should be accuracy and real-time control.Machine vision inspection technology that has features of non-contact and real-time has been widely used in industrial production. Applying machine vision defection technology to control system of flexible roll forming to make real-time detection for sheet in the forming process can realize the real-time and accurate control of forming process. Combining the technology of flexible roll forming feedback control and machine vision inspection and the image analysis technique, this paper presented a mapping algorithm for plate section contour and developed online detection system of flexible roll forming based on machine vision technique by programming with VC++and OpenCV software. In the flexible roll forming process, the image preprocess and analysis are made by periodically collecting the section contour, then the mapping model is model according to the extracted geometric feature and theoretical profile curve of the profile image and the obtained errors are fed back to the control system, realizing the closed loop control. Experimental results show that the system can realize reliable and real-time detection for simple and complex plate cross-section contour and verify the correctness and feasibility of the proposed algorithm which have guiding and refer significances for the online detection development of flexible roll bending forming system based on machine vision technique.