Numerous research papers have already demonstrated the theoretical benefits of matrix-structured assembly systems. Nevertheless, such assembly systems have hardly been used in practice so far. The main reason for this, apart from the technical integration, is the complexity of controlling matrix-structured assembly systems. In theory, decentralized, agent-based control architectures have proven to be particularly suitable. However, order release has been largely neglected so far. Accordingly, the authors' previous work includes a conceptual approach for capacity-oriented order release in matrix-structured assembly systems. This previous approach calculates possible paths of an order and their capacity requirements considering both routing and sequence flexibility. Furthermore, by combining the possible paths of released orders with orders to be released and comparing them with the available capacity, the previously suggested approach can systematically carry out capacity-oriented release decisions. However, the NP-hard (NP: non-deterministic polynomial-time) problem arising from the consideration of all possible paths has a negative impact on the scalability and real-time capability of order release. Therefore, the present paper aims to extend the previously developed approach. By determining the most likely paths that a given order will take through the assembly system, the combination possibilities are limited in such a way that the total amount of calculations required to find a suitable order for release is reduced. Doing so, the NP-hardness of the previously developed approach can be circumvented. This work contributes to the practical realization and economic operation of matrix-structured assembly systems. The paper describes the logic of path prediction in detail and evaluates its impact on order release.
|