Accelerating PSO based feedrate optimization for NURBS toolpaths using parallel computation with OpenMP

Abstract

Over the last few years generation of a time-optimal feedrate profile for CNC machines has recieved significant attention. This is a difficult optimization problem usually requiring long computation time. In the proposed solution, optimization is performed by parallel Particle Swarm Optimization with Augmented Lagrangian constraint handling technique. In order to decrease computation time the authors previously developed algorithm was reimplemented using Open Multi-processing. OpenMP utilizes the ability of modern CPUS to run multiple threads and reduce the algorithm's runtime by using parallel processing. The performance gain (speed-up) of the algorithm parallelized on a multi-core system has been tested. The experimental results of a time-optimal feedrate profile generated using an example toolpath are presented to illustrate the capabilities of parallel computation to improve the algorithm's performance.