| Summary: | 碩士 === 國立中正大學 === 資訊工程研究所 === 101 === This study introduces the minimum latency pickup and delivery problem (MLPDP), which seeks a low-latency route to transport commodities from pickup nodes to delivery nodes and is applicable to time-sensitive services such as farm fresh distribution conveying perishable goods. The objective of the MLPDP takes viewpoint of customer rather than service provider and optimizes customer satisfaction in terms of latency encountered between demanders and the corresponding suppliers. This study takes the composition of the freight into account and considers the average time elapsed aboard over the customer requests as the latency of a delivery node. To tackle the MLPDP, we design a GA with two edge aggregate crossover (EAC) operators, EAC based on available delivery node count (EAC-C), EAC based on contribution to fitness value (EAC-F) and the enhancement strategy called forward insertion (FI). A series of experiments was conducted to verify the effectiveness of the proposed EAC-C, EAC-F and FI on the MLPDP with LIFO loading and FIFO loading. Statistical test reveals that employing either EAC-C or EAC-F achieves significantly lower total latency than GA with order crossover regardless of LIFO loading and FIFO loading. In reference to LIFO loading, involving FI outperforms ordinary GA process in terms of average solution quality and convergence behavior. Additionally, empirical analysis indicates that FI in LIFO loading associates with both high success rate and improvement ratio compared with FIFO loading.
|
|---|
