Learning-Based Accurate Cycle-Time Prediction for CNC Machines

    Advised by Dr. Sencer, Oregon State University, The United States

    This project aims to present a new strategy to accurately predict machining processing time (cycle time) for industrial manipulators, especially computer numerical control machines.

    Conventional strategies cannot accomplish accurate cycle-time estimation due to neglecting interpolator dynamics of numerical control systems. We propose a framework that considers controller-dependent interpolator dynamics based on machine learning techniques to predict cycle time for any tool-paths and machines.

    Experimental results against realistic tool-paths using virtual/actual machines validate its effectiveness in predicting cycle times accurately.