Digital twin-driven intelligent spinning technique for curved surface parts

Abstract

Spinning is an advanced forming technology widely used in manufacturing of curved surface parts in petrochemical, aviation and aerospace industries. Since the spinning is a local loading and incremental forming process, the workpiece forming status and forming rules are both complex and time-varying, which pose great challenges to the precisely control of spinning process. To address this, a novel digital twin-driven (DT-driven) intelligent spinning technique was proposed. It develops a non-contact measuring device to monitor the workpiece forming status. Utilizing both real-time and historical monitoring data, a twin model of forming status evolution is constructed using deep neural networks. In addition, an efficient multi-objective optimization method is established to achieve online dynamic optimization of spinning process. By integrating the above technologies, the developed DT-driven intelligent spinning technique can well capture the real-time workpiece forming status and time-varying forming rules, moreover, intelligently and gradually design the optimal process aligned with the time-varying forming rules throughout the spinning process. This changes the traditional trail-and-error spinning method, which predetermines the entire process by characterizing it as a linear time-invariant process, thus effectively enhancing forming quality, forming efficiency, and environmental sustainability.