高阶非线性微分方程的并行演化建模算法

A Parallel Evolutionary Algorithm on Modeling High Order Differential Equation

文章提出了动态数据的高阶非线性微分方程的并行演化建模算法。动态系统的传统模型多采用线性模型，不能充分描述复杂系统的动态行为，建立高阶非线性微分方程模型更具普遍意义。我们通过将建模过程分解为可以分布并行的“模型结构选择”和“模型参数优化”二个步骤，在异构网络环境中用 CORBA实现了高阶微分方程建模的并行演化算法。实验表明，新的并行演化算法对于复杂动态系统的建模是十分有效的。

This paper proposes a parallel evolutionary algorithm of modeling dynamic systems by higher－ order differential equations(HODEs).The traditional dynamic systems models are mainly linear models, which can reflect very limited dynamic behavior. The non－ linear HODEs models are much more popular.We realize the parallel evolutionary algorithm using distributed CORBA object on the heterogeneous networking by paralleling the two steps of modeling, which are the structure－ selection and parameters－ optimization. The experiments show that the new parallel evolutionary algorithm is efficient.