Container berth scheduling using genetic algorithm

Pao Yue-kong Library Electronic Theses Database

Container berth scheduling using genetic algorithm

 

Author: Chan, Bing-nam
Title: Container berth scheduling using genetic algorithm
Degree: M.Sc.
Year: 1997
Subject: Cargo handling
Genetic algorithms
Hong Kong Polytechnic University -- Dissertations
Department: Multi-disciplinary Studies
Pages: ix, 117 leaves : ill. ; 30 cm.
InnoPac Record: http://library.polyu.edu.hk/record=b1405416
URI: http://theses.lib.polyu.edu.hk/handle/200/5380
Abstract: Container berth scheduling is an important planning function in container terminals. The problem in berth scheduling is to compile daily shipment schedules deciding “what cargo, when, by which crane, at which berth” in an efficient manner. This complex task has traditionally been assigned to experts, but standardization is desired to improved planning performance and allow for job rotation and the eventual retirement of experts. There is currently wide interest in Genetic Algorithm (GA) among the AI community worldwide. The interest is driven by the inadequacy of linear programming and rule-based system in solving complex resource-planning and scheduling problems. GAs are robust in domains where gradient-based algorithms become trapped in local minima. In addition, GAs tend to visit the less-obvious regions of the search space, which is normally not visited by some other search methods. GA is a probabilistic optimization search technique inspired from the idea of natural selection and genetics. Classical GA operates on a population of feasible solutions which are encodes as bitstrings. Fitness of each bitstring is derived from the objective function values of the solutions. Pairs of bitstrings are selected based on fitness and undergo crossover to exchange information among each other. Mutation operator then works on the newly formed bitstrings and changes the bit values of the strings occasionally. It allows the GA to change search direction as in hill climbing techniques. Selection, crossover and mutation carry on until a new generation of population is created. Then the process goes on with evaluation of the fitness of the individual in the population again and repeat until a pre-determined number of generations is reached. In this process, better solutions are expected with better average fitness for each generation. A hybrid approach of GA is adopted in this study due to the potential difficulty of container berthing scheduling and concern in computational complexity. The GA being developed is going to be aided with heuristics. Hard constraints are handled by simple heuristics to narrow down the search space of GA. Soft constraints are modeled as the objective function of the schedules produced by GA, and the GA in turn optimizes the heuristics applied to the vessels.
However, it is intended that the operators of this hybrid GA are developed as generic as possible such that operators of classical GA can be used with little modification. However, integer strings are used instead of bitstrings since they are direct mapping of the heuristics selected. Linear ranking scheme is used instead of scaling since ranking produced good results on multimodal functions in experiments performed by researcher. Like a classical GA, generational replacement scheme is used. A prototype of container berth scheduling system using GA is written in MATLAB. The objective function of the problem is formulated to measure the quality of a feasible schedule. The function is a linear combination of four performance measures: berthing schedule. The function is a linear combination of four performance measures: berthing delay, port time, proportion of container overflow and proportion of violation in berth preference. Weight of each measure is determined by Analytic Hierarchy Process based on data gathered from questionnaires completed by experts in berthing scheduling. The prototype runs on a Pentium-133 PC with 32 Mb RAM and Windows for Workgroups. It can handle scheduling of 27 vessels in 5 berths having 19 quay cranes for a time horizon of one week within 5 minutes, which is very efficient under the hardware configuration. The prototype is evaluated by performing scheduling based on past vessel arrival data. Objective function values of the schedules obtained from the prototype are compared with the value of the actual schedule. The qualities of the schedules obtained from the prototype using hybrid GA, with heuristics playing an important role, is found to be statistically better than that of the schedule obtained with the existing manual methods. In short, the prototype provides promising and cost effective solution, in terms of hardware and the relatively ease of implementation, to computerized container berth scheduling.

Files in this item

Files Size Format
b14054164.pdf 3.385Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

     

Quick Search

Browse

More Information