基于遗传算法的稻米品质质量等级评价

The Application of Real Coding Based Accelerating Genetic Algorithm in Evaluating Rice Quality

针对以往在水稻品种审定过程中稻米品质评价方法存在人为权重赋值误差,以及简便计算等方面的不足的问题,应用一种新的稻米品质质量综合评价方法,对2006年吉林省区试水稻品种品质质量进行评价,为今后有针对性地采取对应措施提高吉林省的稻米品质质量提供参考.根据国家相关稻米品质标准,运用基于遗传算法的投影寻踪等级评价模型(PPE)进行稻米品质质量综合评价.评价结果表明,参加水稻区域试验的18个中早熟水稻品种稻米品质质量整体水平不高,除了5个品种稻米品质等级为2级稻米外,其它品种均为3级或4级,稻米垩白、碱消值和胶稠度指标是影响稻米品质质量的主要因子.各品种稻米品质质量等级的经验值与PPE计算值之间的平均绝对误差仅为0.1325,PPE模型精度较高;改善稻米垩白、碱消值和胶稠度性状应成为今后吉林省稻米品质育种工作重点.

Aiming at solving the jamming of weight matrix and deficiency of simple cal- culation in the evaluation method to rice quality in the approved process of rice varieties, a new comprehensive evaluation method was applied in rice quality assessment, which the rice varieties came from variety regional test of Jilin province in 2006, in order to provide reference for take measures to improve the quality of rice quality. Projection pursuit grade evaluation （or PPE for short） model based on accelerating genetic algorithm was applied to the rice quality evaluation according to the relevant national standards. The results show that the rice quality level of the 18 medium-early maturing varieties was not high in the overall level, rice quality level of five varieties was 2 grade, and the one of other varieties was 3 or 4 grade, the indices of chalkiness, alkali spreading value, gel consistency was the main factor for rice quality. The absolute error between the experience value and value of PPE was only 0.1325 for the rice quality grade and the computational results applied PPE model was more precise and reasonable; Improving the traits of chalkiness, alkali spreading value, gel consistency should become the key work of rice breeding for quality.