Gene Expression Programming In Evolutionary Modeling

People always abstract the complex problems and phenomena to a simple model for their researching with the modeling's important role in the fields of engineering application and science research.And it is impossible to get a mathematic model which can show the inherent laws of the complex systems' observed data.such as meteorological data,oceanographic date,seismic data,and economic data with the traditional artificial modeling method.Through the continuous tries and repeated tests,Evolutionary Modeling can generate automatically the function model having a nice fitting precision with the evolutionary algorithm and a certain searching strategy.For we need only make a decision for the basic items according to the characteristics of the problem without the necessary of determining the structure of the model at first like applying the artificial modeling.Now the popular,algorithms for the evolutionary modeling are GP and GEP.GP directly applies the tree structure as the code,and when dealing with the problem of the complex system modeling,it is easy to form the code dilation leading to a rapid decreasing of the searching efficiency for the indefinite accreting of tree depth.And GEP applying the linear and head pulsing tail structure having a certain length can be applied in evolutionary modeling properly with the simple penetrate operation and the good algorithm stability.This paper discuses the evolutionary modeling problem applying GEP.The main contents are as follows.Firstly,this paper expounds the key technologies of the GEP,analyzes the coding advantages and makes a comparative analysis on the properties of the two popular modeling algorithms------GEP and GP.Secondly,this paper analyzes the principles of the evolutionary modeling with GEP giving two modeling cases and makes a conclusion through analyzing the modeling results that GEP can generate a function model having a better fitting precise.At last,for the deficiency that the fore constants the only 1 in the model applying only the GEP algorithm,this paper improves the GEP algorithm with compounding the GEP algorithm and the GP algorithm to optimize the constants in the model.Making two modeling examples applying the improved algorithm,this paper finds through the comparative analysis of the models results that the model whose structure and constants has been both improved has a better fitting precise and can more reflect the inherent laws and the connections of the complex system datum.