一个基于知识库的棉红铃虫种群动态预测系统

A Prediction System for Population Dynamics of Cotton Pink Bollworm Based on Knowledge Base

根据人工智能领域中专家系统的基本原理,初步建立了棉花红铃虫种群动态预测的知识库系统。知识库由历年红铃虫发生消长和为害、气象因素和棉花生育资料,以及国家“六·五”、“七·五”棉花病虫害综防攻关研究成果和有关专家积累的红铃虫管理实践经验等知识组成。系统以反向推理和不精确推理策略作红铃虫种群的发生期和发生程度的预测。发生程度的预测分5个等级:低、偏低、中、偏高和高。对时间系统预测符合率在95％以上。

According to the basic priniciples of the expert system in the artificial intelligence domain, a knowledge base system was constructed for forecasting the population dynamics of the cotton pink boll-worm. The knowledge base consists of (1) the historical data of climates, cotton growth, population dynamics and infestation of pink bollworms; (2) recent advances in the integrated pest management of cotton; (3) expert experiences in managing the pink bollworm population. The population dynamics of the cotton pink bollworm is predicted by using backward reasoning and inaccurate strategies. The results of running the system can be expressed in two ways, (1)the period of occurrence of the insect, and (2) the extent of the population which is divided into 5 classes (very low, low, moderate, high, and very high). The accuracy rate of prediction by the system is over 95% in the whole season of cotton growth.