基于混合演化的自适应建模及其应用

针对传统的遗传规划方法收敛效率低、模型精度不高等缺点,将进化策略应用于模型参数的全局最优搜索,在遗传规划建模的基础上使用进化策略修正模型参数,以实现对模型结构和参数的同时优化,通过2种演化算法的结合提高建模精度,同时采用最优保持策略加速优化过程,改善算法的收敛效率.将该方法应用于某型船用汽轮机组的仿真建模中,通过和传统演化建模的对比,证明采用混合演化策略的建模方法具有更快的求解速度和更好的全局搜索能力.

Adaptive hybrid evolutionary modeling method and its application

An improved genetic programming modeling method has been proposed to improve the traditional genetic programming method's shortcomings: low constringent efficiency and non high model precision. This method used evolution strategy to modify the parameters of models which established by means of the genetic programming. By the combining of GP and ES, the global optimal searches of model parameters and structures were carried out at the some time, which heightened the modeling precision. By simultaneously adopted optimum keeping strategy, we also ensured algorithm's constringent efficiency. Used this method to modeling and simulated a certain type marine steam turbine and compared the simulative results with general GP method, we find the hybrid evolutionary modeling method has faster modeling speed and better global search abilities.