拖轮动态调度的混合演化策略算法设计

Design of hybrid evolutionary strategy algorithm of dynamic tugboat scheduling problem

针对单向航道港口,采用基于时间窗口的滚动调度策略对拖轮作业动态调度问题进行研究。以拖轮最大作业时间最小化为目标,采用滚动时序的方法构建动态调度模型。利用基于3种不同局部搜索策略的混合演化策略算法对该模型进行优化求解,设计个体修正方法,保证所有个体在优化过程中均为可行解。算例分析结果表明,该策略在收敛速度和求解结果方面较传统粒子群算法、模拟退火算法及基本演化策略算法更优,在3种不同的局部搜索策略中,基于互换操作的局部搜索策略对算法性能的优化更加理想。通过一系列对比实验对算法的参数设置进行分析研究。

The rolling scheduling strategy based on time window was adopted to research the dynamic tugboat scheduling for the one-way channel port.The scheduling model was established using the method of rolling time series with the objective of minimizing the tugboats＇ maximum make span.A hybrid evolutionary strategy algorithm based on three different local search strategies was designed to optimize the model,and an individual correction method was designed to guarantee all individuals were feasible in the process of optimization.The results verified by the example analysis show that the convergence speed and solution results of hybrid evolutionary strategy algorithm are better than that of the traditional particle swarm optimization,simulated annealing algorithm and basic evolutionary strategy algorithm,and the performance of the algorithm is optimized more ideally when local search strategy based on swap operations is adopted in three different local search strategies.Parameter settings of the algorithm were analyzed through a series of contrast test researches.