Search

Manufacturing companies typically use sophisticated production planning systems optimizing production steps, often delivering near-optimal solutions. As a downside for delivering a near-optimal schedule, planning systems have high computational demands resulting in hours of computation. Under normal circumstances this is not issue if there is enough buffer time before implementation of the schedule (e.g. at night for the next day). However, in case of unexpected disruptions such as delayed part deliveries or defectively manufactured goods, the planned schedule may become invalid and swift replanning becomes necessary. Such immediate replanning is unsuited for existing optimal planners due to the computational requirements. This paper proposes a novel solution that can effectively and efficiently perform replanning in case of different types of disruptions using an existing plan. The approach is based on the idea to adhere to the existing schedule as much as possible, adapting it based on limited local changes. For that purpose an agent-based scheduling mechanism has been devised, in which agents represent materials and production sites and use local optimization techniques and negotiations to generate an adapted (sufficient, but non-optimal) schedule. The approach has been evaluated using real production data from Huawei, showing that efficient schedules are produced in short time. The system has been implemented as proof of concept and is currently reimplemented and transferred to a production system based on the Jadex agent platform. [ABSTRACT FROM AUTHOR]

Temporal difference learning and eligibility traces are two mechanisms for solving reinforcement learning problems. The temporal difference technique bootstraps the state value or state-action value at every step as with dynamic programming, and learns by sampling episodes from experience as in the Monte Carlo approach. Eligibility traces is a mechanism that offers a means for recording the degree of which state is eligible for undergoing learning process. This paper aims to investigate the underlying mechanism of eligibility traces strategies using on-policy and off-policy learning algorithms. In doing so, the performance metrics can be obtained by defining the learning problem in a simulation environment, in conjunction with different learning algorithms. However, measuring learning performance and analysing sensibility are very expensive because such performance metrics can only be obtained by running an experiment with different parameter values. This paper proposes a comparative study for analysing the mechanism of eligibility traces. The objective of this paper is to compare and investigate the influences on performance caused by those different approaches. [ABSTRACT FROM AUTHOR]