The Application Of Parallel Genetic Algorithm On Heat Conduction Inverse Problem

With developments of technology, computing power and speed of new generation computers are much better than the former ones. However people's demands for high performance computing are increasing and infinite in some sense, so that in addition to enhancing the computing power of a processor, parallel processing is also an efficient way to enhance the computing power of a system. In the past, the parallel processing can only be performed on the expensive and special computers. Along with the cost of PCs and network descending, the distributed parallel computing concept is widely used in the parallel computing. This thesis focuses on determination of parameters in a two dimension heat conduction equation on the MPI network parallel environment, by solving an inverse problem using the parallel Genetic Algorithm.Firstly this thesis introduces research background and significances related with the subject, then shows basic theory of the parallel computing, introduces the cluster concept and the message passing system of MPI, after that, discusses how to establishes a parallel computing environment based on the MPI and Linux; and then based on network background, integrates principle of parallel algorithm with parallel characteristic of parallel genetic algorithm (PGA), using the PGA to solve the thermophysical properties inverse problem of ceramic/metal combine material, The PGA combines high-speed parallel-ability of supercomputers with the inherent parallelity of GA, and improves greatly the efficiency and accuracy of GA s. Among master-slave, fine-grained and coarse grained parallel approaches, the coarse-grained model is most widely used for its little communication overhead and its diversifying of the population. This thesis studies the influence to performance of parallel algorithm of difference migration step and crossover points of PGA, analyzes and compares some experiment results. Finally, conclusions of this thesis and suggestions for further research are given.This thesis has six chapters.Chapter 1 introduces research background and significances related with thesubject and main work which has been done.Chapter 2 introduces development and the classification of the parallel computer, discusses the basic theory of parallel computing.Chapter 3 discusses the GA and its parallelity, introduces its theory background and algorithm description.Chapter 4 introduces the MPI system; a detail configuration of MPI cluster system which this thesis has used is given. The techniques of MPI parallel programming also have been addressed.Chapter 5 gives the detail solution process of thermophysical properties inverse problem. The influence to performance of parallel algorithm of difference migration step and crossover points of PGA has been analyzed.Chapter 6 summarizes this thesis, and suggestions for future research are given.This subject is supported by National Natural Science Foundation of China (NSFC Grant No. 60173046).