基于主成分分析的BP神经网络在水华预测中的应用

Application of BP Neural Network Based on Principal Component Analysis in Algal Bloom Prediction

摘要: 随着淡水生态系统水体污染和富营养化进程的加剧，诱发了大面积水华，其不仅破坏生态系统，而且造成巨大的经济损失。因此，根据水体各个理化因子对水华的发生进行预测就显得尤为重要。首先根据池塘1~15周的数据，利用主成分分析法对影响浮游生物总量的13个理化因子进行主要影响因子分析，得到池塘水华发生的主要影响因子为：总氮、透明度、溶解氧、铵态氮、盐度、总磷，溶氧。其次，根据确定的7个主要理化因子作为BP神经网络的输入层，浮游生物量作为输出层来对水华发生进行预测。结果表明基于主成分分析的BP神经网络模型的预测结果与真实值的拟合系数高达0.9912。为此，本文的研究方法可有效地预测水华的发生。

Abstract: With the intensification of water pollution and eutrophication of freshwater ecosystems, large areas of algal bloom have been included, which not only destroy the ecosystems, but also cause huge economic losses. Therefore, it is very important to predict the occurrence of algal bloom according to the physical and chemical factors of water body. Firstly, according to the data of the pond for 1~15 weeks, the main influencing factors of 13 physical and chemical factors affecting the total plankton were analyzed based on principal component analysis (PCA). The main influencing factors of algal blooms were total nitrogen, transparency, dissolved oxygen, ammonium nitrogen, salinity, total phosphorus and dissolved oxygen. Secondly, according to the main seven physical and chemical factors identified as the input layer of BP neural network, the plankton biomass was used as the output layer to predict the occurrence of algal bloom. The results show that the fitting coefficient between the predicted result and the true value of the BP neural network model based on principal component analysis is as high as 0.9912. Therefore, the research method in this paper can effectively predict the occurrence of algal bloom.