In this research, neural and artificial network and support vector machine have been used to evaluate the wastewater of Khorramshahr treatment plant. Also, the possibility of using the city's sewage for agricultural purposes was investigated. In this study, the monthly
More
In this research, neural and artificial network and support vector machine have been used to evaluate the wastewater of Khorramshahr treatment plant. Also, the possibility of using the city's sewage for agricultural purposes was investigated. In this study, the monthly values of BOD, COD, TS and TSS that were required in this study were used and also these values were used to evaluate the wastewater of Khorramshahr treatment plant for use in agriculture. The application of artificial neural network model is also possible to predict the quality of effluent from wastewater treatment plants. The selected ANN (LM) model had good accuracy in estimating BOD5. However, this model had poorer performance in predicting maximum values. Using a two-stage network search optimization algorithm, the optimal values of the characteristics of the SVM model, namely ɛ, C and γ, were obtained as 0.037, 13 and 1.472, respectively. Finally, according to the results obtained in this study, the SVM model was recommended for BOD5 time prediction for Khorramshahr refinery. According to the results obtained from the qualitative analysis of treated wastewater, effluent from BOD5 removal is equal to 88%, COD is equal to 92%, TDS is equal to 70% and TSS removal is equal to 27%. Khorramshahr wastewater treatment plant effluent salinity was measured with a minimum salinity of 208, maximum 3050 and an average of 1544 micromoles per centimeter. Therefore, the effluent of the city's wastewater treatment plant is in group C3, acceptable water. Based on the amount of sodium in the effluent of the treatment plant for irrigation of wheat, barley, soybeans, figs, olives, poplar and the like, according to the Wilcox diagram, there are no restrictions on the use of this effluent.
Manuscript profile