Self-aware and self-correcting machine tools for robust accuracy

点击翻译按钮获取中文摘要

Much wealth is created by machining complex parts for the aerospace, automotive, oil and medical industry to name a few. The ability to make ever more complex parts, with tighter tolerances, enables original equipment manufacturers (OEMs) to design products that stay ahead of international competitors in terms of performance and cost. It also enables part manufacturers to competitively supply Canadian and international OEMs. In both cases, such industries are users of machine tools often widely available. What differentiates users is their capacity to select and then get the most out of their installed manufacturing base. In recent years, a growing trend has been towards machines which combine processes such as turning and milling to produce complex parts in a single setup thus potentially increasing productivity and quality. But their high capital cost means that uptime is crucial and making an out-of-tolerance part is unacceptable even for the first part made. Accuracy and uptime of machine tools demand a fundamental understanding of the machine error sources, their tracking in the industrial environment (self-awareness), their compensation (self-correction) and the anticipation of malfunctions or excessive loss of accuracy. Beside the basic geometric construction deviations, thermal effects, elastic deformations and wear are also known to occur but are not understood in the context of multi-axis machine tools. In an industrial setting, the ability to measure such errors and distinguish them requires rich mathematical models. This is because directly measuring those errors individually on a frequent basis is too time consuming and not always possible since errors sources act simultaneously. Instead, indirect approaches are favoured whereby their combined effect as volumetric errors are measured using non-intrusive in-situ methods and then error separation techniques applied. So, at a fundamental level this research program proposes an original holistic approach integrating new geometric, thermal, elastic and wear models of multi-axis machine tools developed using painstaking laboratory techniques. Then, indirect approaches are explored for the estimation of the parameters of such models for a particular installed machine in industry. The sought industrially viable data gathering techniques will favour an original approach based on scanning probes and uncalibrated brought-in and indigenous artefacts, the latter consisting of features already available in the machining volume, to gather data about the machine instantaneous status. The ease with which data will be made available will enable the timely detection of trends in machine behaviour. The rich calibrated models form the basis for real time machine compensation and detection and anticipation of excessive deviations so that corrective actions can be planned by machine users.