No abstract

A logistics-based project is described which addresses the need for better intermodal transport, whilst balancing economic and environmental gains through the use of Internet technologies. Pipeline intermodal system to support control, expedition and scheduling (PISCES) provides an integrating platform for using these technologies in processing and sharing commercially sensitive data within transport chains (i.e. road, rail and barge). The paper demonstrates how information from an Internet-based system can be used to drive a scheduling tool to provide appropriate routes for the transport of goods, using a multimodal transport model.

The combination of Petri nets (PNs) as an analysis tool for discrete-event dynamic systems and artificial intelligence heuristic search has been shown to be a promising way to solve flexible manufacturing systems (FMS) scheduling problems. However, the NP hard nature of the problem obscures the PN capability of reasoning about the behavior of the system. In this paper, two techniques to alleviate this drawback are presented: a systematic method to avoid the generation of futile paths within the search graph and a novel hybrid stage-search algorithm. The new algorithm is based on the application of guided by a PN-based heuristic within a limited local search frame. An optimization policy is applied to maintain, under evaluation, only the most promising paths. For each system state, the algorithm is able to decide whether an enabled operation should be applied and to maintain this decision until new information forces reconsideration. This eliminates permutation paths and useless scheduling sequences. Experimental results show that the algorithm's cost does not grow exponentially with the size of the problem. Comparison with previous work is given to show the superiority of our approach and the potential of PN-based heuristic search.

The combination of Petri net (PN) and AI to solve flexible manufacturing systems (FMS) scheduling problems has been proven to be a promising approach. However, the NP-hard nature of the problem prevents the PN capability of reasoning about the behavior of a practical system. To overcome this drawback, we propose two techniques: a systematic method to avoid the generation of unpromising paths within the search graph and a stage-search based algorithm. The algorithm developed is based in the application of the A* algorithm and the PN-based heuristics. The search is performed within a limited local search window where an optimization policy is applied to evaluate the most promising paths. For each state, the algorithm is able to decide whether an enabled operation is applied, and to maintain the decision until new system information makes the reconsideration meaningfirl. Comparison with previous work is presented to show the superiority of the proposed approach.

No abstract