【摘要】 针对传统的BP神经网络应用于藻类生长预测时,往往出现训练时间较长、输出数据精度低等问题。本文提出了在含有两层隐含层的BP神经网络结构中,对于数量一定的神经元,若神经元在隐含层分配合理,则BP神经网络可以达到减少训练次数并且能满足问题精度的要求。应用实例表明,该方法对预测藻类生长显得非常有效。更多还原

【Abstract】 Traditional methods based on BP neural network for forecasting alga’s growth has the problems of training time is too long, and data’s precision is not enough. This paper gives a method on how to distribute a certain number of neuron to a BP neural network’s two hidden layers. This method can reduce BP neural network’s training time and can satisfy the data’s precision. Experiments are done to explain and verify this method.更多还原