Research On Intelligent Stowage Decision On Full Route For Container Ship

With the containerization and enlargement of container ships in maritime transport,both container terminal throughput and individual vessel carrying capacity have significantly increased.Intelligent container ship stowage decision-making plays a vital role in ensuring ship safety and improving operational efficiency.In order to address the difficulties of existing planning models being inapplicable or suboptimal modeled for the full route container loading problem,this paper takes "Intelligent Container Vessel Stowage Planning for Full Routes" as the topic and conducts systematic theoretical exploration and research.A full route stowage plan decision-making model that considers the reverse container volume,special container types,and multiple ship statics constraints is envisaged,and a resolution algorithm which takes into account the multimodal nature of the objective function and the diversity of the decision space is designed.The main contributions in this paper are summarized as follows:(1)Prediction of Container Ship Voyage Stowage Volume.A ship voyage container volume prediction method based on a hybrid kernel function Support Vector Machine(SVM)is proposed to address the problem of delayed and lacking stowage information of subsequent ports in container ship routing decision-making.The method determines the correlation between influencing factors and the predicted sequence utilizing the theory of grey relational analysis,measuring the differences of various influencing factors and assigning different weights to them.The parameters of the SVM prediction model are optimized by using the genetic algorithm,which improves the prediction performance and robustness of the model.The prediction model proposed in this paper has a relatively low minimum error rates of-0.11%,which has better prediction performance compared with the single kernel function model,and better generalization ability than the BP neural network model.(2)The Container Ship Stowage Master Bay Planning Problem.Building upon the prediction of voyage container volume,this paper investigates the master bay planning problem in the two-stage container ship stowage planning over entire route.A Mixed Integer Linear Programming(MILP)model is constructed in this paper in order to achieve the minimum volume of container rehandles of full route and reduce additional fuel consumption caused by the longitudinal inclination of the vessel,while also considering practical constraints such as the overall longitudinal strength,local strength,rib shear and bending moment,hatch cover influence,dangerous goods containers,and the mixed loading of different container types simultaneously.A two-stage heuristic algorithm is designed aiming to solve the MILP model within a limited time,in which the first stage obtains the lower bound of the total number of container rehandles for the entire route,providing a feasible initial solution for the search of the second stage heuristic algorithm,and realizing the solution of the MILP model within a limited time.An experiment designed to generate a loading plan for a 20,000 TEU vessel is carried out,and the results show that the MILP loading plan based on the two-stage heuristic algorithm is solved within minutes,providing a feasible plan and a relatively short solving time.(3)The Container Stowage Problem Within Bays.Based on the output of the master bay planning as input,this paper studies the container stowage problem within bays in the two-stage stowage planning for container ships over the entire route.The main-beam stowage Integer Programming(IP)model is constructed,considering the stowage of refrigerated containers,high-cube containers,dangerous goods containers,and mixed containers of different sizes.A Greedy Randomized Adaptive Search Procedure(GRASP)algorithm combining preconceived construction and randomization strategies has been designed,which is incorporated with loading rules and human experience aiming to enhance its ability to solve IP models.Experimental verification is conducted on both publicly tested instances and instances generated in this article,indicating that the proposed IP model found higher bottom limit and lower top limit,and the GRASP algorithm achieves fast solving times in seconds without the need for parameter adjustment.Considering the specificity of refrigerated and dangerous goods containers,the feasibility of the proposed model for practical operation is enhanced.(4)Relaxation of container ship loading constraints.It is studied how multi-objective evolutionary algorithms can solve the global container intelligent loading problem over the entire route,and tighten the relaxation phenomenon of the two-stage decomposition method from a global perspective.A high-dimensional multi-objective container loading decision model is constructed taking the practical constraints such as ship structure and yard stacking layout into account.Besides,an improved algorithm based on the local search component of the Non-dominated Sorting Genetic Algorithms(NSGA)III is designed to solve this model.Compact chromosome encoding technique is adopted to represent feasible solutions for container stowage,using ternary arrays to represent each gene of the chromosome,which avoids the problem of unsolvable stowage models when dealing with a large number of decision variables.This paper investigates a chromosome repair technique to avoid generating invalid chromosomes,and develops neighborhood operators to improve the performance of mutation operator and local search.Performance testing on various instances shows that the NSGA-III algorithm based on local search is feasible and effective in solving high-dimensional multi-objective container stowage planning problems,and outperforms NSGA-II and Random Weighted Genetic Algorithm.All experiments in this paper are implemented using Matlab programming language,and the mathematical programming model is solved by CPLEX solver.The experimental results,which simulate real stowage cases and verify the generated stowage plans for actual ships,demonstrate the effectiveness of the proposed intelligent container stowage decision-making approach.This research is of practical significance for improving maritime safety and promoting the development of intelligent ship equipment,and lays an significant theoretical foundation for achieving navigational efficiency.