Hybrid Genetic Algorithm For Bin Packing

Three dimensional bin packing problem is a loading problem which puts many small objects into big objects(such as containers) reasonably. Container loading problem belongs to the three dimensional bin packing problem and the space source optimization problem with several constraints. Because packing optimization can increase container's utilization ratio and decrease transport cost, the corporation profit is improved. As well, the research results can be applied to space source optimization problems such as plane hold loading, ship hold loading, even the distribution of computer memory.Bin packing problem is a NP-complete problem. Because its complexity, approximate algorithms is used in practice and it is difficult to solve it precisely. And precise solution is unnecessary in many situations the point generally lies in how to come up with a satisfactory solution as soon as possible.It begins by summarizing the bin packing problem in sorts and their research method, as well as analysis on the current research on the bin packing problem. The main work lies on applying genetic algorithms to solve the bin packing problems. And based it, in order to improve the utilization ratio and shorten the computation time, an efficient approach for bin packing which is genetic algorithm combined with heuristic method is indicated. An intelligence loading system employing Delphi language is developed on the basis of the previous hybrid algorithm, which is propitious to general bin packing problem. Finally, part of the running results is analyzed.Genetic Algorithm is a random method for searching based on the mechanics of natural selection and the process of evolution. Its core problem is to search the coordination between efficiency and stabilization of optimization problem, which is called robustness. And it is more proper to deal with the complex and non-linear problems that are difficult for traditional search methods.In this dissertation, the basic principle of genetic algorithms as well as the design...