A Novel Integrated Approach to the Enhanced Production of Polyhydrobutyrate with Mixed Culture in Activated Sludge

Polyhydrobutyrate (PHB) production from fatty acids by activated sludge is mainly influenced by the ratio of initial substrate concentration (S-0) to initial biomass concentration (X-0), ratio of chemical oxygen demand to nitrogen (COD/N), and aeration strength. In this work the PHB production by activated sludge was evaluated and enhanced using an integration of the response surface methodology (RSM) and a novel machine learning method, that is, support vector regression (SVR) coupled with the accelerating genetic algorithm (AGA). The results demonstrated that the SVR-AGA model was superior over the RSM model. With the SVR-AGA approach, a maximum PHB yield of 185.48 nig PHB/g VSS under the optimum conditions of S-0/X-0 of 0.286, COD/N of 144, and aeration intensity 0.004 m(3)/(L . h) was estimated. The verification experiment confirmed that the measured optimum PHB yield was close to that estimated by using the SVR AGA approach. This work indicates that an integration of the SVR and AGA methods was an appropriate and effective approach for evaluating and enhancing the PHB production process by mixed cultures like activated sludge.