Precision digital twin of production machines for sustained high accuracy and traceable part conformity

点击翻译按钮获取中文摘要

In many manufacturing processes a machine moves a tool relative to a workpiece. Depending on the process, such as subtractive or additive, the tool can be, for example, a cutting tool, laser beam or deposition nozzle. The geometric and dimensional quality of manufactured parts depends largely on the accuracy of positioning and path following of these machines. Customised products and small batch production increase the demands for highly accurate machines. Accuracy is affected by the built geometry of the machine, which differs from nominal, exhibiting imperfections in the alignment between moving axes and in the motion of each axis. Further geometry changes are caused by wear and tear, thermal effects, compliance or lack of stiffness, and hysteresis. Much resources are needed to make good parts and also to verify their conformity by metrology. Over past decades, much effort has been dedicated to quasi-static geometric errors occurring under no load conditions. This proposal is directed at phenomenon that exhibit non-linear behaviour and/or defy the rigid body hypothesis, which as a result will not be assumed to apply. The research program aims at developing the models, technologies and methodologies to provide a continuously updated live digital twin of the machine to obtain robust and sustained machine accuracy by continuously feeding the machine controller with either direct axis command corrections or updated parameters for in-situ computer numerical controller (CNC) correction models. In parallel, such models could, through future research, issue a traceable conformity status of the produced part thus reducing the need for post-production inspection. In addition to rich models, lean, robust, fast and automated shop-floor compatible means will be researched to match the model to its actual physical counterpart. The digital twin will use both physics-based models, for insight, and error representation models. Detailed experiments using laboratory equipment such as laser interferometers and dynamometers will provide the essential knowledge for model development. Then, means to calibrate the models in-situ using online sensors such as contact and non-contact probes and inertial measuring units, but also sensorless approaches, will be researched and tested. Signals made available by modern machines' CNCs will be exploited fully for this purpose. The program will advance the understanding of machine tools non-linear and non-rigid error sources. In addition to a scientific culture, the program will provide HQPs with expertise in manufacturing metrology, machine errors and their compensation. Canadian industry will have early access to the new knowledge and technologies before their international competitors through technology transfer and through the trained HQPs.