Benchmarking of Methods and Instruments for Self-Optimization in Future Production Systems

Abstract

Technologically demanding products are manufactured by adoption of modern production technology. The flexibility required for the ambitious technological processes needs a new kind of controlling mechanisms, which can only be reached by sophisticated optimization approaches like Self-Optimization. For Self-Optimization different approaches for controlling technologies are available, especially tools using cognitive information processing techniques. These new technologies have to be evaluated concerning important performance indicators against the background of the production process characteristics. Aim of the benchmarking process in this context is to ensure model quality of models used in a cognitive software application that is presented in this paper. As an example, an optimization approach using a combination of Artificial Neural Networks and a Soar optimizer is presented, with a benchmarking example of Artificial Neural Networks modeling process parameters and product characteristics resulting.