道路货运量预测的遗传规划方法研究

Genetic Programming Method of Road Freight Forecast

影响道路货运量预测的因素 ,往往呈复杂的非线性关系。文章探讨了遗传规划方法在解决该类问题上的应用 ,从随机生成的初始经验公式出发 ,运用复制、交换、变异等进化方式进行不断优化 ,最终得到能真正、全面描述问题实质的经验公式。以青海省历年货运量及经济发展状况为样本 ,采用遗传规划方法自动找出货运量随时间变化的规律 ,并对未来货运量进行了预测。该方法避免了事先确定变量之间函数关系的主观性 ,从而使预测结果更加客观、有效。

Many factors of road freight forecast have complex and nonlinear relations. This paper discusses the application of genetic programming method to solving this kin d of question. Starting from the initial experiential formulas generated randoml y and using the genetic programming method, this paper optimizes the structures of these experiential formulas by the evolution ways of reproduction, crossover or mutation, and finally finds the best experiential formula, which can complete ly describe the essence of the problem. Taking the road freight and economic dat a of Qinghai province as an example, the changing rules of road freight with tim e are found by the genetic programming method. Then, the road freight of the com ing year is forecast. This method avoids the subjectivity of deciding the functi on relation among variables in advance. The forecast result is objective and val id by using the genetic programming method.