Research of Tugboat Scheduling Optimization Based on Q-Learning and Genetic Algorithm

Tugboat scheduling plays a crucial role in modern ports, as it directly affects the arrival and departure of vessels. This study focuses on optimizing tugboat scheduling in multi-berth bases. By analyzing the operational process of port tugboats and the characteristics of scheduling, a mathematical model is developed to dynamically optimize tugboat scheduling with the objective of minimizing completion time. To achieve this, a tugboat scheduling optimization method is proposed, based on Q-learning and genetic algorithm within the framework of pattern-driven scheduling. The solution to the problem is represented by chromosomes with priority rule encoding. These chromosomes are divided into gene patterns, serving as the state in the staged scheduling algorithm. Additionally, a heuristic reward function is constructed in the state space of the learning system. By using performance prediction variables, this function guides and expedites the search process of the genetic algorithm. Furthermore, the search process achieves a balance between exploitation and exploration by optimizing the genetic operator and the relevant parameters of Q-learning. The proposed tugboat scheduling method is validated through simulation experiments to demonstrate its effectiveness.