Parallel implementation of metaheuristics for optimizing tool path computation on CNC machining

Abstract

The incorporation of technological advances in industry is a must, even for traditional sectors where most companies are SMEs and investments are limited. Technology can be used to increase productivity and the quality of the manufactured product. Drilling is a common procedure in industry. It usually consists of multiple drilling of a flat surface with a tool. Usually the tool is placed on the surface to be drilled at a safe distance and then it makes the drilling in a linear fashion. Optimization of the tool path often involves reducing the movement of the tool to place it over the next point to be drilled, known as airtime. The problem of minimizing airtime for drill paths is highly complex. Most proposals to solve the problem try to adapt it to the formulation of the Traveling Salesman Problem (TSP), in which the objective is to navigate a list of nodes using the minimum global distance. In this paper, the purpose is to provide a solution to the TSP applied to tool path optimization by means of a Discrete version of the Teacher-Learner-Based Optimization (TLBO) algorithm. To improve performance, the algorithm is implemented using a parallel Computer Unified Device Architecture (CUDA) and run on a manycore Graphical Processing Unit (GPU). The results show that the parallel implementation of Discrete TLBO is faster than 9x the sequential implementation.