中文摘要：

针对作物产量预测，提出基于商空间粒度计算的分析法。在商空间粒度计算理论思想下，分析作物产量序列中粒度的选取，用属性划分方法对论域x进行颗粒化，对属性f取不同的粒度进行颗粒化。通过属性的粒度变化对论域进行划分，得到新的商空间并应用其解决问题，可以降低问题复杂度。通过商空间理论中的分层与合成技术选取大小合适的粒度，能全面获取产量序列中的信息，也更加符合人类智能特点。冬小麦产量预测实验结果也证明这种粒度分析和选取方法是有效的。

英文摘要：

The granular analytical method of crop yield forecasting is put forward. How to select an appropriate granularity is analyzed in the crop yield series based on the theory of quotient space and granular computing. New quotient space can be obtained after universe X granulation by means of attribute partition and attributefgranulation by adopting different granularity. And the new space can make complex problem simple. Appropriate granularity can be selected by using delamination and synthesis method in the quotient space theory. It can get more accurate information in the yield series and be more close to human intelligence. Experimental results of wheat yield forecasting also show that the granular analytical method is more effective.