TY - GEN
T1 - Production planning and control via service-oriented simulation integration platform
AU - Xu, Dong
AU - Son, Young Jun
PY - 2014
Y1 - 2014
N2 - In this paper, a service-oriented simulation integration platform is proposed to support manufacturing production planning and control of a complex manufacturing system. In particular, a multi-level service composition structure is considered, where simulation models at different levels of the hierarchy (e.g. equipment, shop, and enterprise) can be seamlessly and efficiently integrated. The Service oriented architecture Modeling Language (SoaML) is then employed to specify the service capabilities, service interfaces, service data model and chorography related to production planning and control. Furthermore, the proposed approach is demonstrated through single-period and multi-period inventory management. For the single-period inventory control, the optimal product price is estimated under different demand variability. For the multi-period inventory control, the convergence of a multi-agent reinforcement learning algorithm is demonstrated considering the eligibility trace. The proposed platform has been successfully deployed for integrating various different simulation models (e.g. discrete-event, agent-based, systems dynamics, process simulation). In addition, experiments illustrate the impact of demand variability on the product price, and the learning results of the optimal decision policy.
AB - In this paper, a service-oriented simulation integration platform is proposed to support manufacturing production planning and control of a complex manufacturing system. In particular, a multi-level service composition structure is considered, where simulation models at different levels of the hierarchy (e.g. equipment, shop, and enterprise) can be seamlessly and efficiently integrated. The Service oriented architecture Modeling Language (SoaML) is then employed to specify the service capabilities, service interfaces, service data model and chorography related to production planning and control. Furthermore, the proposed approach is demonstrated through single-period and multi-period inventory management. For the single-period inventory control, the optimal product price is estimated under different demand variability. For the multi-period inventory control, the convergence of a multi-agent reinforcement learning algorithm is demonstrated considering the eligibility trace. The proposed platform has been successfully deployed for integrating various different simulation models (e.g. discrete-event, agent-based, systems dynamics, process simulation). In addition, experiments illustrate the impact of demand variability on the product price, and the learning results of the optimal decision policy.
KW - Inventory control
KW - Reinforcement learning
KW - Service-oriented architecture
KW - Simulation
UR - http://www.scopus.com/inward/record.url?scp=84910050599&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84910050599&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:84910050599
T3 - IIE Annual Conference and Expo 2014
SP - 351
EP - 360
BT - IIE Annual Conference and Expo 2014
PB - Institute of Industrial Engineers
T2 - IIE Annual Conference and Expo 2014
Y2 - 31 May 2014 through 3 June 2014
ER -