Distributed Multi-Objective Cooperative Coevolution Algorithm for Big-Data-Enabled Vessel Schedule Recovery Problem
Fatemeh Cheraghchi, Ibrahim Y. Abualhaol, Rafael Falcón, Rami Abielmona, Bijan Raahemi, Emil M. Petriu · 2020
During a maritime voyage, delays due to disruptive events could result in financial and reputation losses. The vessel schedule recovery problem (VSRP) aims at adjusting vessel speeds to mitigate the negative impact of such delays. The granulated speed-based vessel schedule recovery problem (G-S-VSRP) is a big-data-enabled VSRP. It is a multiobjective optimization problem defined by dividing the trajectory between ports into regions (encoded by geohashed system) and mining the speed profiles in these regions from Automatic Identification System (AIS) data. The G-S-VSRP minimizes delay and financial loss of a vessel voyage; it also maximizes the speed compliance with the historical navigational patterns. Using geohash-based speed mining on AIS data in the G-S-VSRP gives rise to a large-scale optimization problem, where the number of speed variables in geohashed regions grows to the order of thousands. Due to the complexity of such a problem, traditional multiobjective evolutionary algorithms (MOEAs) would stop improving or showing steady behavior. We improve the MOEA's performance using a cooperative coevolution algorithm based on a divide-and-conquer approach to deal with large-scale optimization problems. We introduce a novel Distributed Multiobjective Cooperative Coevolutionary Algorithm (DMOCCA) to improve the performance of MOEAs.