Artificial neural network model of algae density in Xiangxi Bay of Three Gorges Reservoir

Jun-Yu Liu, Hua-Jun Luo and Yi

Abstract

Algae density model of diatom and blue algae in Xiangxi Bay of Three Gorges Reservoir was established by artificial neural network (ANN) technology. Using stepwise multiple linear regression method, the important environmental factors (nitrate, ammonium nitrogen, total phosphates, phosphate, silicate and water temperature) were selected as input variables in ANN model. The optimal structure of the ANN model was determined, which has two hidden layers (number of neurons in the first hidden layer: 5; number of neurons in the second hidden layer: 4). The ANN model has been shown to perform well for simulating the algal density of diatom and blue algae simultaneously, in which the training set R2 values are 0.738 (diatom) and 0.949 (blue algae), the test set R2 values are 0.773 (diatom) and 0.870 (blue algae), respectively.