KL4SiM II – Automated Generation of Control Strategy Variants in Simulation Models of Production and Logistics Using Combinatory Logic

• Applicants: Professor Dr. Jakob Rehof; Professorin Dr.-Ing. Sigrid Wenzel
• Subject Area: Production Systems, Operations Management, Quality Management and Factory Planning
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 511349842

In the continuation research project, a technology for the automated synthesis of control strategy variants for discrete-event simulation models in production and logistics will be developed. Building on the predecessor project KL4SiM, which enables the generation of structural variants, this project extends the approach to include the rule-based composition of control strategy variants with combinators and predicates. Combinators represent building blocks for the synthesis of complex control strategies from simpler components, which can be assembled into strategies with the help of combinatorial logic. The aim of the research project is the automatic generation of control strategy variants in order to investigate the behavior of the simulation model structure variants under the systematic variation of control strategies and thus improve the overall range and quality of simulation experiment results. To achieve this goal, a taxonomy for control strategies of production and logistics systems is first developed, which serves to systematize the variance of control strategies as well as the consideration of interdependencies between model structures and control strategies. Based on the taxonomy for control strategies, tool-independent combinators are designed and formalized in a type system for the synthesis of control strategy variants. This type system ensures that only syntactically and semantically correct control strategies are synthesized. The formalism for structural variants from KL4SiM is extended and the existing synthesis framework CLSP (Combinatory Logic with Predicates) is adapted accordingly. A central challenge is the automatic configuration of the synthesis and the definition of suitable predicates to consider the dependencies and constraints between the generated structures and control strategies. Comprehensive experiments are designed and executed for the empirical investigation of the developed synthesis technology. For this purpose, a simulation-based experiment field will be developed that enables the systematic investigation of synthesized model variants under variation of control strategies. In addition, a criteria-based evaluation methodology is being developed that considers both software-related evaluation criteria and criteria relating to the effort involved and the quality of the results. This ensures that the synthesized control strategy variants are not only formally correct, but also practically applicable and efficient. At the end of the project, an evaluated technology for the generation of control strategy variants and a procedure for the application of control strategy variant synthesis will be available.

• DFG Programme: Research Grants