| 초록 |
There can be a bunch of measurements in a process, for instance of activated sludge process, BOD, COD, temperature, SVI, DO, MLSS, turbidity, color, etc. Among the measurements some are easily measurable while the others are not, e.g. BOD needs 5 days while DO for every minute. Multivariate regression method is favorable candidate to overcome the time mismatch. If there is a relation between the readily and hardly measurements, combination of the handies can be used to predict the nuisances. This research suggests a probabilistic method for the regression in which holds two critical concepts, i.e. the latent variable to represents condition of the process, and the probabilistic reasoning to interpret the regression results. Combining them enable engineers to analyze the process by substitution the headaches for handies. |
