[目的]秋季低温冷害对晚稻生产不利影响极大,准确预警其出现对夺取晩稻高产稳产意义重大。[方法]釆用模糊数学知识原理研制田阳秋季低温冷害始现时段预警输出模型,选配适合于模糊事件秋季低温冷害初日出现时段的隶属函数,计算各预报因子隶属度,结合数理统计学知识,建立多元线性回归预警输出模型,作秋季低温冷害初日出现时段第一次预警；第二次预警,采用模糊数学并、交运算逻辑关系原理制作预警模型。[结果]第一次预警历史拟合率87.3%,第二次预警历史拟合率88.5%。试警四年,第一次试警正确三年,正确率75.0%,二次试警一年,试警正确。[结论]秋季低温冷害模糊数学预警输出模型,技术科学,方法易学易懂,历史正确拟合率、试警准确率结果理想。

Low temperature and cold damage in autumn has great adverse effect on late rice production, accurate early warning is of great significance for the high and stable yield of late rice. Using the principle of fuzzy mathematics, we studied the starting time early warning output model of low temperature and cold damage in Tianyang, selected a membership function suitable for the initial date emergence period of the low temperature and cold damage in autumn, calculated the membership degree of each prediction factor, built the multiple linear regression with the knowledge of mathematical statistics, and made the first early-warning of initial date emergence period of the low temperature and cold damage in autumn. Then, we used the principle of fuzzy mathematics and the logical relation between parallel and intersection operations to make the second early-warning. The results showed that the historical fitting rate of the first early-warning reached 87.3%, the historical fitting rate of the second early-warning reached 88.5%. During four years’test early-warning, the first early-warning was correct in three years, the correct rate reached 75.0%. The second early-warning was tested for one year, and the result was correct. The fuzzy mathematics warning output model for low temperature and cold damage in autumn is scientific and easy to understand, the correct rate of historical fitting and the accuracy rate of test early-warning are ideal.