Research on online prediction of deformation of thin-walled parts based on digital twin technology

Thin-walled parts are widely used in the aerospace field. They have the characteristics of large material removal during processing and are affected by factors such as processing technology and processing environment, which can easily cause the deformation of parts. Deformation will seriously affect the machining accuracy of parts, which is one of the key factors affecting the installation accuracy and quality of parts. Aiming at the problem that the machining deformation of thin-walled parts cannot be perceived in real-time during the milling process. This paper constructs a digital twin framework for the milling process of thin-walled parts. Unity3D and Visual Studio 2010 software are used to build a digital twin intelligent processing platform to realize the interaction and integration of physical objects and virtual models. The simulation is carried out by ABAQUS finite element analysis software and verified by experiments. At the same time, based on the simulation data and historical processing data, the PSO-LSTM thin-walled part deformation prediction model is constructed, and the prediction model is embedded into the intelligent processing platform. The real-time monitoring of deformation during the machining process of thin-walled parts is realized, and the problem that the deformation of thin-walled parts cannot be perceived in real-time during the machining process is solved. By processing Al-7075-T6 thin-walled parts for experimental analysis, the prediction accuracy reaches 96.1%, which verifies the accuracy and effectiveness of the platform.