西北地区小麦黄矮病流行分析及人工神经网络模型的构建

Analysis of the wheat yellow dwarf epidemic and its predicting model of BP neural network in northwestern China

对西北地区半干旱气候区小麦黄矮病1992--2009年发生、流行情况进行长期监测、分析，选择制约小麦黄矮病发生、流行的23个因素，利用三层人工神经网络可以逼近任意连续函数，对非线性预测系统进行模拟处理的特点，分析所选预测分子，提出一套完整的建立BP人工神经网络模型的方法，并建立陕西省BP神经网络长期预测模型。对1992-2006年数据进行网络训练，利用2007-2009年数据进行测试。结果表明，以发病率为指标，输出结果误差在0．001～0．034之间；以发病级别作为预测结果，模型计算得出的数值与实际病级完全吻合，准确率为100％。说明利用神经网络建立小麦黄矮病预测模型是可行的。

We select 23 factors as restrict elements for the studies of occurrence and epidemic of wheat yellow dwarf from 1992 to 2009 in semi-arid climate area of northwestern China. Using the function of three-layer artificial neural network can approximate arbitrary continuous with arbitrary precision and simulated by nonlinear forecast system. We put a method of BP neural network model and set up the model through selecting 23 factors based on the data of 1992 to 2006. Prediction of 2007 to 2009 has been carried out. Results showed that the error ratio is between 0. 001 and 0. 034 when using incidence as the output results. Based on the predicted result indicating by disease grade, the output number of model is fully consistent with actual disease level, and the accuracy was 100%. The result also proved a feasible method of prediction wheat yellow dwarf use the BP neural network.