基于遗传算法的投影寻踪模型对小花碱茅最佳收获期的判定

针对小花碱茅实际生产中最佳收获期难于客观确定的问题，提出基于加速遗传算法的投影寻踪模型用于解决小花碱茅（Puccine Uiatenuiflora）最佳收获期的判定问题，该模型通过优化指标投影方向，将多维数据指标（样本评价指标）转换到低维子空间，并根据投影函数值的大小评价出各样本的优劣，最大限度地避免了权重矩阵取值的人为干扰。结果表明：初花期收获各指标投影函数值最大，这个时期收获的牧草具有相对较高的营养价值和产量，并具有相对较低的粗纤维含量，因而这个时期收获综合评价值最高。图1，表2，参18。

Evaluation on the Optimal Harvest Time of Puccinellia tenuiflora Based on RAGA-PP Model

In order to ascertain the optimal harvest time of Puccinellia tenuiflora in practical production scientifically, PPC (Projection Pursuit Classification) model based on RAGA (Real coding based Accelerating Genetic Algorithm) was applied. Projection direction was optimized and high-dimension data were converted into low-dimension space with PPC, and the value of projection indexes was obtained. Based on these projec-tion function values, the sequence of samples from the big to the small was gained. At the same time, subjective impact on weight matrix farthest could be avoided with PPC model. The results showed that the initial bloom stage have the biggest value of projection indexes, and its comprehen-sive evaluation is the best. Puccinellia tenuiflora harvested in the initial bloom stage had higher nutritive value, higher output and lower coarse fiber content. This model offered a new method to solve the evaluating problem of the optimal harvest time of Puccinellia tenuiflora.