Research On The Optimization Model And Algorithm Of Outbound Container Storage Location Assignment Problem Of One Bay In Container Terminal

Container terminals in China are facing huge challenges as they develop rapidly.On the one hand,currently affected by the coronavirus pneumonia,the number of outbound containers in China has been significantly decreased,the loading and unloading efficiency of the storage yard in various container ports low,and the vessel stay time in container ports extended.On the other hand,large vessels require container ports to load and unload a large amount of outbound containers within a relatively small time window(several days),which not only causes greater pressure on the production activities of ports in the short term,but puts forward higher requirements on outbound container storage location assignment(OCSLA).Therefore,on the premise of the existing resources and equipments of the container terminal,how to quickly load and unload outbound containers at the storage yard to minimize the number of reshuffles and the operating costs caused by OCSLA has become one of the hot issues facing in container terminals.However,the container terminal is a comprehensive and complex system.There are many factors involved in solving outbound container storage location assignment problem(OCSLAP),such as storage strategies and ship arrivals.And information completeness often exists when relevant information is obtained.Hence,these issues that how to make the model reflect the actual situation of the storage yard,how to stack outbound containers to match the ship stowage plan when the information is complete,and how to solve OCSLAP to minimize the number of reshuffles when the information is unceratain,need further in-depth analysis and discussion.Driven by the trend of port digitalization,many ports have accumulated large amounts of actual data of outbound containers,but the utilization rate of data is low,and the analysis and decision-making are also poor.Consequently,it is of great practical significance to conduct the research based on actual data to solve OCSLAP and it has also become a hotspot of container port research.In this context,how to use the real data to innovatively predict the weight levels and the arrival sequences of outbound containers and how to deal with the optimization problems of container storage and pre-marshalling before shipment,so as to improve the quality of OCSLA and reduce the vessel stay time in port,are another important subject that need further study.As a consequence,OCSLAP is studied and the main research contents are summarized as follows:(1)The OCSLAP model is studied from the perspective of the system.The container terminal is a highly complex system,and each link and element affects each other.Dividing port-related issues into a series of sub-problems for the specific decision-making is an effective way to solve port problems.Therefore,on the basis of analyzing the overall layout of the container terminal and the loading and unloading equipments,the relationship between the storage yard planning,berth allocation,stowage planning,and resource allocation from a systematic perspective is studied,and then the influencing factors of OCSLAP is analyzed.Afterwards,the relevant factors involved in OCSLAP under complete and incomplete information are determined and constraints set,decision variables set and objectives set are set up.Finally it forms an architecture that solves OCSLAP and is regarded as the theoretical basis for specific decision models in subsequent chapters.(2)The OCSLAP model under the condition of complete information is established and the heuristic algorithm based on stacking rules is proposed.In the case that the arrival orders and weights information of outbound containers are complete,a 0-1 integer programming model is established to minimize the number of reshuffles by considering constraints such as the stacking rules and storage status of outbound containers.To solve this model,three stacking principles,i.e.,the least reshuffle rule,lowest stack rule and nearest stack rule,are extracted from the literature research and the port practice survey and five heuristic algorithms,i.e.,LRR?LSR?NSR,LRR?NSR,LRR?LSR,LSR?LRR?NSR and LSR?LRR,are developed.Finally,the superiority of the proposed algorithm is proved through method comparison,and the number of reshuffles and the reshuffle rate at different bay scales are determined through numerical analysis,which provided an effective decision basis for port managers to choose appropriate locations.(3)The OCSLAP model is established under the condition of uncertainty,and a two-stage heuristic algorithm is designed.On the basis of the OCSLAP model under complete information conditions,a two-stage stochastic programming model is devised with the objective of the minimum number of reshuffles by considering the impact of incomplete information on the arrival sequences of outbound containers.Different information levels have different effects on the results.To this end,the vessel incomplete information is is divided into little,moderate and high levels.To solve the model,a two-stage heuristic algorithm is designed.In the first stage,CPLEX is used for outbound containers whose arrival orders are determined.In the second stage,five proposed heuristic algorithms are applied.Finally,the number of reshuffles and the reshuffle rate under different information levels are determined through numerical analysis,which proves the effectiveness and practicability of the algorithm.(4)Combined with the actual data of the arrival orders and weights of outbound containers,the OCSLAP under data-driven conditions is studied.The port actual data is used to capture the correlation law and the obtained results can better serve the port,thereby improving the loading efficiency of vessels.However,the results of the practical survey in Guangdong-Hong Kong-Macao Greater Bay Area show that container pre-marshalling operations are also an important factor influencing the loading of outbound containers.Therefore,it is of great significance to study the OCSLAP in the two stages of stacking and pre-marshalling operations from the perspective of actual data for accelerating the loading speed of ships and reducing the vessel stay time in port.This section mainly solves three problems: first,the weight levels are predicted by using machine learning methods,and the arrival orders are determined by using discrete Markov chain prediction method,which provide basic data for building the mathematical model;Second,on the premise that the initial stacking status is known,the effective relocation rules are extracted from literature research and practical investigation,and the heuristic algorithm is designed to solve container pre-marshalling problem(CPMP);Finally,the specific cases are analyzed from the weight attributes,destination port attributes,and multi-attribute three different perspectives to determine the optimal ratio of the number ofoutbound containers that normally and urgently enter the storage yard in different circumstances and provide a decision-making basis for port staff to reasonably arrange the number of outbound containers that enter the storage yard.This subject is closely integrated with the storage yard management practice.The research results help to improve the operation level of the entire terminal and provide a basis for port staff to reasonably arrange the locations of outbound containers.At the same time,it also provides theoretical and methodological references for the research on container port related dispatch.The proposed heuristic algorithms in this article have important reference significance for the solution of similar optimization problems such as the storage of automatic warehouses.The information incompleteness and data-driven situation involved in the study are aimed at solving the complexity problems of the port,which have certain guiding significance for the actual operation of the container terminal.