Abstract: In order to improve the accuracy of forecasting grain production, for the BP neural network prediction of grain production fall into local optimum easily, based on the concentration of the immune system regulation mechanism and Genetic Algorithm global optimization characteristics, an Adaptive Immune Genetic Algorithm optimization（AIGA） is presented to optimize BP neural network’s initial weights. And a specific optimization process is given. The optimum neural network is used to realize simulation prediction of grain yield. The results show that compared with BP neural network and genetic neural network forecasting method, this optimized network model can obtain more accurate results for predicting grain yield.

Key words: BP neural network, self-adaptive Immune Genetic Algorithm, grain prediction

摘要： 为了提高预测粮食产量的准确度，针对BP神经网络进行粮食产量预测时易陷入局部最优的缺陷，主要借鉴免疫系统的浓度调节机制和遗传算法的全局寻优特性，用自适应免疫遗传算法（AIGA）来优化BP神经网络的权值和阈值，并给出了具体的优化过程。用优化的神经网络对粮食产量进行了仿真预测，通过仿真实验表明，与BP神经网络预测法和遗传神经网络预测法对比，优化的网络模型在粮食产量预测中取得了更精确的结果。

关键词: BP神经网络, 自适应免疫遗传算法, 粮食预测