Abstract:

A BP neural network of ultrasonicassisted biodegradation of EE2 and BPA was established, the amount of carbon source, amount of nitrogen source, inoculum dose, ultrasonic time, and degradation time were set as input, and the degradation rates of ethinyl estradiol (EE2) and bisphenol A (BPA) were set as output. The BP neural network model was employed for predicting the response for the factorial design of ultrasonicassisted biodegradation of EE2 and BPA. The primary effects and interaction effect on the ultrasonicassisted biodegradation were analyzed by factorial design. The optimum conditions for EE2 and BPA degradation were obtained. The correlation coefficient between experimental and predictive values were 0.952 5 and 0.983 1 for EE2 and BPA, Nash Suttclife coefficient values were 0.956 5 and 0.957 2, which indicated the accuracy prediction of the models. The results of factorial design analysis suggest that the amount of carbon source×the amount of nitrogen source, and the amount of carbon source × inoculum dose both had a synergistic effect on the degradation of EE2 and BPA. The maximum degradation rates of EE2 and BPA arrived at 87.13% and 69.27%, respectively. The optimum conditions for the degradation were as follow: 10% of carbon source, 10% of nitrogen source, 20 mL of inoculum dose, 1 min of ultrasonic time, and 168 h of degradation time. The biodegradation difference analysis shows the degradation rate of EE2 and BPA were proportional to mobility, while inversely proportional to the persistence.

Key words: ethinyl estradiol(EE2), bisphenol A(BPA), BP neural network, factorial design, ultrasonicassisted degradation, Pseudomonas putida