Predicting biofilm thickness and biofilm viability based on the concentration of carbon-nitrogen-phosphorus by support vector regression

Lin, Shanshan; Wang, Xinmin; Chao, Yunlong; He, Yude; Liu, Ming

Predicting biofilm thickness and biofilm viability based on the concentration of carbon-nitrogen-phosphorus by support vector regression

2016

Lin, Shanshan | Wang, Xinmin | Chao, Yunlong | He, Yude | Liu, Ming

Current tools to predict biofilm thickness and viability in spatial distribution are poor, especially those based on chemical oxygen demand (COD), total nitrogen (TN), and total phosphate (TP) due to their limited data and complex calculations. Here, support vector regression (SVR) was used to predict biofilm thickness and viability in a reactor filled with carriers of crushed stone globular aggregates. Analyses combined confocal laser scanning microscopy and flow cytometry with Kriging interpolation revealed that biofilm thickness varied from 22 to 31 μm, and biofilm viability decreased from 80 to 30 % in the flow direction of the reactor. The biofilm thickness at the bottom was thicker than that in the upper layer, but biofilm viability contrasted with biofilm thickness in the vertical distribution. The values of biofilm thickness and viability were predicted at a layer 35 cm from the bottom of the reactor with mean squared error values of 0.014 and 0.011, respectively. Correlation coefficients were 0.996 and 0.997 between carbon-nitrogen-phosphorus (C–N–P) removal with biofilm thickness and viability in spatial distribution, respectively. This study provided an important mathematical method to predict biofilm thickness and viability in spatial distribution based on the concentration of C–N–P.

Показать больше [+]