摘要
针对传统方法解决动态系统微分方程建模问题所遇到的困难和存在的不足,设计将方程进行串结构编码并用进化方法进行演化建模的算法,以串形结构表示结构,用进化算法优化结构和参数,成功地实现了动态系统的常微分方程组建模过程的自动化。计算实例表明:采用此算法能够在极短的时间内由计算机自动发现多个较优的常微分方程组模型,与原来GA和GP结合的方法相比较,它具有建模过程智能化、模型结构非常灵活多样、数据拟合和预测精度更高等优点。
To overcome the difficulties and the draw backs in modeling dynamic systems by traditional methods, a hybrid evolutionary modeling algorithm （HEMA） was proposed to model the Dynamic system with a system of ordinary differential equations （ODEs） whose main idea was to embed string coded idea and evolution algorithm optimize system. The model structure was shown by string structure, and the evolution algorithm was used to optimize the structure and parameters of a model. It succeeded in automating the modeling process of systems of ODEs for dynamic systems. The result of experiments indicates that, by running the HEMA, the computer can search out multiple superior ODEs models within a reasonable time. Compared with the combination of a genetic algo- rithm （GA）, which was employed to optimize the parameters of a model, and a genetic programming method, which was employed to optimize a model structure, this algorithm, possesses advantages, such as a more intelligent modeling process, more flexible and various model structures, higher precision of fitting and predicting data.
出处
《山东大学学报（理学版）》
CAS
CSCD
北大核心
2008年第11期27-30,35,共5页
Journal of Shandong University(Natural Science)
基金
国家自然科学基金资助项目(60573065)
山东省自然科学基金资助项目(Y2007G33)
山东省重点学科基金资助项目
关键词
动态系统
串结构
进化算法
常微分方程组
dynamic systems
string structure
evolutionary algorithm
system of ordinary differential equations
