Estimation of nitrate reductase enzyme parameters in activated sludge using an extended Kalman filter algorithm

This paper demonstrates an application of the extended Kalman filter to a wastewater plant using real process data to estimate parameters in a semi-mechanistic model. This technique for parameter identification allows a semi-mechanistic model developed for pure cultures to be applied to a mixed culture population where isolation of enzyme kinetic parameters is not practical. This paper demonstrates an application in which a wastewater treatment facility with three physical unit operations and an unusual operating profile is modeled as 11 reactor units with recirculation streams. On-line nutrient probe data from two locations in the aerobic reactor are combined with operational data for flows and feed composition. This paper demonstrates the successful application of a biochemical model developed at the bench scale to this facility through the use of an extended Kalman filter. The estimated biological model parameters and the covariance matrix diagonal converge to stable values, indicating a successful implementation.