SOM神经网络在农业旱情评价中的应用

Application of SOM Neural Network to Comprehensive Evaluation of Agricultural Drought

在分析现有旱情评价方法的基础上,根据农业区域旱情评价指标体系和等级划分标准,利用MATLAB仿真软件建立了自组织映射SOM神经网络区域旱情综合评价模型,并对1997年关中西部的冯家山、宝鸡峡、石头河、羊毛湾4大灌区进行了旱情评价。结果表明:1997年关中西部4大灌区与重度干旱的聚类接近,说明4大灌区在1997年都发生了重度干旱,这与相关文献的评价结果相同,也与陕西省历史资料记载基本吻合。

Based on the analysis of existed evaluation method to drought,according to evaluation system and rating standard of agricultural drought,a self-organizing map neural network（SOM） model was established using MATLAB simulation software,and the model was applied to drought comprehensive evaluation in four large-scale irrigation districts of Fengjiashan,Baojixia,Shitouhe,Yangmaowan in the western Guanzhong in 1997.The results show that the four large-scale irrigation districts have undergone severe drought in the year,the evaluation result is the same with other relevant literature and the Shaanxi historical drought records.