基于粗糙集和BP神经网络的粮食产量预测研究

Study on prediction of grain yield based on rough set and BP neural network

为提高粮食产量的预测精度,提出一种基于粗糙集和BP神经网络的粮食产量预测方法。该方法以吉林省粮食总产量的历史数据作为研究对象,利用粗糙集理论的属性约简特性,识别与粮食产量相关性较大的影响因素,剔除非主要影响因素,利用约简后数据建立RSBP神经网络预测模型。结果表明,粗糙集理论能有效减少数据的维数及噪声,减少神经网络的计算量,结合两种方法能有效提高预测速度和精度。

In order to improve the accuracy of the prediction of grain yield, a prediction method of grain yield based on rough set and neural network BP is proposed. The historical data of the total grain yield of Jilin Province was taken as the research object, it used the characteristic of the attribute reduction to identify the factors associated with grain yield correlation of rough set theory, and to eliminate the secondary influence factors. Then the prediction model of RSBP neural network was established by using it. The results was shown that rough set theory could effectively reduce the dimension and noise of data, and reduced the amount of neural network computations. The combination of 2 methods could effectively improve the prediction of speed and precision.