Research On A Dynamic Multi-Objective Knapsack Problem And Its Algorithms

Many optimization problems in the real world usually have some complex characters,such as multi-objective,dynamic and etc.These complex characteristics posed a great challenge to the traditional optimization algorithms since they may not be able to obtain the satisfactory results.Due to the strong adaptability of the natural selection and genetic evolution mechanism,evolutionary algorithms have been widely used to solve the multi-objective or dynamic optimization problems by a lot of researchers.The corresponding research topics have also been termed as evolutionary multi-objective optimization or evolutionary dynamic optimization.It is noticeable that the existing research works have considered only one character.However,many optimization problems in the real world have always involved multiple complex characters simultaneously.In recent years,the researchers in the community of evolutionary computation have begun to concern in those complex optimization problems that have multiple complex characters,especially dynamic multi-objective optimization problems.Therefore,this thesis will study and investigate evolutionary algorithms for dynamic multi-objective knapsack problems on the test benchmark generator and algorithm design.The methodologies of different disciplines,such as systems engineering,control engineering,operations research,computation science and applied mathematics,are integrated into the framework of this thesis.The main research works include the following three parts:(1)to investigate the current research situation in evolutionary dynamic multi-objective optimization through reviewing the related works on the traditional multi-objective optimization theory,evolutionary dynamic optimization and dynamic multi-objective optimization problems;(2)to propose a new generator for dynamic multi-objective knapsack problems based on XOR operator through investigat-ing the research works on multi-objective knapsack problems and dynamic optimization benchmark generator;(3)to introduce three different strategies into the well-known multi-objective evolutionary algorithms,that is,NSGA-? and MOEA/D,in order to enable these algorithms to solve the dynamic multi-objective knapsack problems,and then a series of randomly-generated test cases will be used to examine the performance of the proposed algorithms in the experiments.Base on the above research works,it can be seen that the proposed generator of dynamic multi-objective knapsack problems can construct the robust test cases and the proposed strategies exhibit the good adaptability in the different dynamic characters of environments.