Driven by innovations in production techniques and tools, factories are becoming more and more flexible. In addition, the growth of technologies such as the Industrial Internet of Things is making production systems holding more and more decisional nodes and entities. Thus, one of the key activities of production management is the efficient scheduling of production tasks. In addition to being a complex combinatorial problem to solve, the nature of the environment makes dynamic scheduling a very challenging problem. This paper addresses the problem of dynamic scheduling of a flexible manufacturing system (FMS), with constraints such as family-dependent setup times and interoperability. To this purpose, the proposed approach combines a set of scheduling rules optimized by an optimization-simulation mechanism. The experiments are performed on two sets of scenarios describing the dynamic arrival of products in the system.