초록 close

The activated sludge method was initially developed in Manchester, England, and has been modified. Also, the reactions in the activated sludge had been studied, and were consolidated with the ASM No. 1(Activated sludge model No. 1) in 1987. This model deals with most parts of the activated sludge process. But, it is hard to say that the ASM No. 1 is easy or correct because it has many parameters and it is difficult to estimate correct values. If we have correct and easy method to analyze them, it'll be a good thing. The activated sludge process is the method using the microbe's growth. So, Important factors in the process are a influent COD, DO, and MLSS related to the microbes' growth. We analyzed those factors with the transfer-ARIMA model, and estimated the effluent COD. The models have been used by daily data and decided four times every season. The forecast is accomplished by them in the following month. The modeling procedures are composed of three steps; identification, estimation, and forecast. The cross-correlation analyses in the identification step is used to test factor's validity. Proper parameters in the estimation step are selected by AIC(Akaike information criterion), SBC(Schwartz's Bayesian criterion), and chi-square test and estimated by the least square method. In the forecasting step, the computed values by the models were compared with the real values using the correlation coefficient, the standard deviation, and MAE(Means of absolute errors). The result compared the computed values with real values was as the followings; the correlation coefficient was ranged from 0.73 to 0.84, MAE was ranged from 4.8% to 7.3%, the standard deviation was from 0.58 to 0.98. As the result of simulating the following month, the correlation coefficient was raged from 0.73 to 0.85, MAE was ranged from 3.8 to 8.6%, and the standard deviation was ranged from 0.42 to 0.65.