通过对偃师市 1 984～ 1 995年棉铃虫历史资料的分析研究 ,运用多因子线性回归方法筛选出相关因子 x1( 6月下旬平均气温 ) ,x2 ( 7月上旬平均气温 ) ,x3 ( 7月上旬相对湿度 ) ,x4 ( 6月下旬降雨量 ) ,并利用模糊综合决策理论建立预测模型。预报质量技术优劣评估 S值检验 ,求和型、加权型和综合决策型 S值均为 0 .9394,历史拟合率均达 91 .67%。实际应用中 ,预测 1 996、1 997、1 998年三代棉铃虫的发生程度与实际情况完全相符 Based on the analysis of the historical data of cotton bollworm from1994 to1995 in Yanshi, the correlation coefficients x1 (average temperature in late June), x2 (average temperature in early July), x3 (7) were screened by using the multi-factor linear regression method Monthly relative humidity), x4 (late June rainfall), and the use of fuzzy comprehensive decision-making theory to establish a predictive model. The S-value test, summation type, weighted type and comprehensive decision-making type S value of forecasting quality technical evaluation are both 0. 9394, and the historical fitting rate reaches 91.67%. In practice, it is predicted that the occurrence of three generations of cotton bollworm in 1996, 1997 and 1998 exactly matches the actual situation