摘要
基于人工神经网络 (ANN)理论 ,面向稻米品质的综合评价问题 ,分别开发了 SOM和 BP神经网模型 ;研究了模型的设计、利用观测数据建立网络结构训练样本集以及网络学习等问题 ;从不同角度分别对杂交籼稻雄性不育保持系和恢复系两组供试亲本的品质样本进行聚类和综合评价 .实际仿真结果表明 ,ANN用于米质评价是科学、有效的 ,而且方法简便。
Based on the artificial neural networks principles, confronting the problem of synthetic evaluation of rice quality, the SOM and BP neural networks models were respectively developed. Further more, this paper discussed the problems of designing and learning neural networks and the way to organize a group of measured data into a sample set for model training. A group of rice quality data were got, they were not only used for classification by SOM, but also for synthetic evaluation by BP. Computer simulation demonstrated that the identification methods were scientific and effective.
出处
《福建农林大学学报(自然科学版)》
CSCD
北大核心
2002年第2期150-154,共5页
Journal of Fujian Agriculture and Forestry University:Natural Science Edition