Options
2015
Conference Paper
Titel
Self-optimization for complex multi-dimensional optimization tasks in manufacturing scheduling
Abstract
Manufacturing systems no longer have to pursue one, but a whole set of goals. Classic optimization regarding lead time or capacity utilization is still sought after, but needs to be extended by factors such as energy consumption or use of cooling lubricants to meet today's requirements. Thus the models of dependencies and system behavior become more complex, hampering optimization by classic algorithmic approaches. One subdomain of the Cluster of Excellence ""Integrative Production Technology for High-Wage Countries"" thus examines the potential of cognitive self-optimization as a way of handling the associated complexity. This paper gives a short introduction into the constraints and dependencies that have to be considered to find overall optima for process chains in manufacturing and describes the associated mathematical complexity. Afterwards, the suitability of different classical algorithmic approaches for multi-dimensional optimization is researched and compared to the concept of self-optimizing process chains. These considerations are supported by extensive simulation runs in Matlab to compare performance criteria such as run time behavior and goal achievement.