Quality embedded intelligent remanufacturing

This thesis is motivated from the four keywords: remanufacturing, quality, multi-agent and intelligence. Recent years' environmental problems caused tightening the regulations and legislations for used products. Therefore remanufacturing is getting more attention. The quality of used products is uncertain and even dynamically changes during the remanufacturing process, and each used product should be individually handled in a different way depending on its quality. Fortunately recent developing wireless technologies like radio frequency identification (RFID) may enable remanufacturing control systems to identify, track, and control each used product and disassembled subassembly/part (PDSP) automatically. The multi-agent approach can be a good solution for the individual control of each PDSP, because a centralized control system is not eligible to managing so many elements in the remanufacturing system. The objective of this thesis is to propose a quality embedded remanufacturing system (QRS) which comprises a multi-agent framework and a scheduling mechanism. First, this thesis discusses the fundamental concepts for the proposed modeling tools and scheduling mechanism: the QRS quality characteristics and the multi-agent framework. As the second step, this thesis proposes QRS modeling tools which support the PDSP/resource quality representation and comprise: intuitive remanufacturing system representation (IRSR) and dynamic token two-level colored Petri-nets (DTPN). The former is designed from the user-side perspective and the latter is from the system-side perspective. The multi-agent framework is constructed based on the model represented with the proposed tools. Last, this thesis proposes a real-time scheduling mechanism for the QRS which enables the constructed framework to execute. The scheduling mechanism embeds a communication protocol among agents and dispatching rules formulated depending on the PDSP/resource quality. A knowledge-based approach is adopted to increase efficiency of the scheduling mechanism, where the knowledge is learned by simulations. A heuristic method is also proposed to reduce the simulation time.