UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Prediction of cycle time using finite impulse response filter and polynomial trajectory generation techniques Tam, Sharon

Abstract

The demand in methods to manufacture parts with complex geometries accurately while minimizing the machining times have driven the technological development of computer numerical control (CNC) machine interpolator algorithms. However, the nature of these algorithms are hidden within the proprietary software of commercial CNC systems, making it difficult to predict the cycle time (i.e. machining time) of a part. This thesis presents the use of finite impulse response (FIR) filtering and polynomial interpolation methods in cycle time prediction algorithms for CNC machines. CNCs use either FIR filter or polynomial based motion trajectory generators to deliver smooth reference position commands to the machine tool’s feed drives. A FIR filter trajectory generation algorithm was developed using two cascaded first-order FIR filters with local corner smoothing by controlling the timing of velocity pulses for 5-axis CNC machine tools. Through the identification of FIR filter parameters from CNC machines using a simple linear toolpath, a cycle time prediction algorithm was developed from the trajectory generation algorithm suited for part programs with longer path segments. The FIR filter-based trajectory generation was experimentally validated to predict cycle times with less than 10% error. Another cycle time prediction algorithm was developed using a polynomial-based trajectory generation algorithm with a cubic acceleration motion profile and spline corner smoothing. The motion parameters such as corner feed, acceleration, and jerk required for polynomial-based interpolation methods are identified for a CNC machine using a corner variance test and approximated using exponential functions. The parameters are fed into the trajectory generation algorithm to predict the cycle time for a part program. The polynomial-based cycle time prediction was proven to predict cycle times within 1% of the actual measured time on the machine.

Item Media

Item Citations and Data

Rights

Attribution-NonCommercial-NoDerivatives 4.0 International