A genetic algorithm for a self-learning parameterization of an aerodynamic part feeding system for high-speed assembly

The aerodynamic feeding technology developed at the IFA allows feeding rates up to 800 parts per minute while maintaining high reliability and variant flexibility. The machine's setup procedure requires the adaptation of only four machine parameters. Currently, optimal parameter configurations need to be identified manually. This task is greatly time-consuming and requires a high level of expertise.

Prospectively, the machine should utilize an algorithm that autonomously identifies optimal parameter configurations for new workpieces to realize fast setup procedures. This paper presents a genetic algorithm for a self-learning feeding system that has been validated in comprehensive simulation studies.