K-L绝对信息散度投影寻踪分类模型及其应用

K-L absolute information divergence of projection pursuit classification model and the application in irrigation engineering

鉴于节水灌溉方案评价中各指标的不确定性,为了解决各单项节水灌溉方案指标评价结果的不相容问题,提高节水灌溉方案评价模型的分辨率,提出了一种有效和通用的模型——K-L绝对信息散度投影寻踪分类模型。采用基于实数编码的加速遗传算法寻找最优的投影方向,同时用最佳投影方向信息研究了各评价指标对节水灌溉方案的影响程度,取得了符合客观实际的评价结果,为节水灌溉方案评价研究提供了决策依据。

In view of the uncertainty of indexes for evaluating water-saving irrigation program,in order to resolve the incompatibility of the evaluation results about water-saving irrigation program,and to raise the resolution of the evaluation model for water-saving irrigation program,we suggested an effective and general model,which is the classification model by K-L absolute information divergence of projection pursuit.It adopts Accelerating Genetic Algorithm based on Real Coding to find the best projective direction,and uses the datum of the best projective direction to research the influence level of each factor on the water-saving irrigation program;the classification results which accord with the fact are obtained,and it provides a decision basis for the water-saving irrigation program.