基于加速遗传算法的组合预测方法研究

A Research on Combination Forecasting Approach Based on Accelerating Genetic Algorithm

应用加速遗传算法解决组合预测问题 ,可以克服传统遗传算法的缺点 :对搜索空间 (优化变量空间 )的大小变化适应能力差 ,计算量大 ,易出现早熟收敛 ,控制参数的设置技术无明确准则指导等 ,与已有结果相比 ,对预测误差信息矩阵无正定性要求 ,目标函数可以推广到规模庞大 。

It can win though traditional genetic algorithm shortcoming applying accelerating genetic algorithm in solving combination forecasting problems: the latter's adapt ability was soft in search space (i.e. optimize variable space), measure quantity big, convergence prematurity, setting technique of controls parameter no definitude instruct rule, etc. The new approach was not need canonicity in forecasting error information matrix, the objective function scale may hugeness, the forecasting precision high ,etc.