Genetic algorithm for bending process in sheet metal industry

Abstract

In sheet metal industries, the required shape has to be bent from the flat 2D sheets. In this bending process, the most complex process is due to the determination of the bend sequence and assigning proper tools for each bend. Finding the best bending sequence is itself a combination problem and this when coupled with tool selection results in a combination explosion, which clearly states that an exhaustive approach is impossible to solve the problem in a time-economic way. Moreover since human operators handle the above operation, a process that is not complex, their desire have to be considered. A genetic algorithm is used to generate the bend sequence and tool selection for each bend. It finds a feasible solution based on collision detection, a simple process and the operator's desire. We present genetic algorithms for finding the best bending sequence and tool selection for each bent, based on collision avoidance and the operators desire, we discuss our new approach towards solving it, and the ideas being implemented and are evaluated using real part data.