浮游植物密度的人工神经网络预测研究

The Study on the Prediction of Artificial Neural Network for Phytoplankton's Density

将人工神经网络模型应用于藻类密度数据的预测计算之中 ,并利用遗传算法对其网络结构进行优化计算以保证计算结果的准确性且自动确定网络结构 ,分别对神经网络的“当天模型”和“预测模型”进行了计算。结果显示 :几种人工神经网络模型在计算精度以及预测数据的趋势上都有较好的效果 ,目前国内学者使用的人工神经网络“当天模型”无法对其后数据进行预测 ,不能起到实际预测的作用 ,而经过遗传算法优化后的人工神经网络“预测模型”不仅达到了很好的预测效果 ,而且网络结构简单 ,适用于浮游植物密度的预测计算。

The “same-day model” and the “predictive model” of artificial neural network is used to predict the density of the phytoplankton in this thesis, and genetic algorithm is applied to optimize the net frame to give the number of the units in the hidden layer automatically. The result shows that the precision and the trend of the predicted data of the artificial neural network models are all satisfactory. The “same-day model”, which is used by the domestic scholars at present, cannot compute the subsequent data, so it's not valuable practically. The optimized “predictive model”, which is not only precise in the prediction but also simple in the frame, is more suitable for the prediction of the density of the phytoplankton.