Quantifying Uncertainties for Chlorine Residual Control in Drinking Water Distribution Systems

摘要

An approach to modelling for robust model predictive control of chlorine residuals in drinking water distribution systems (DWDS) was proposed in previous paper. A model for the next control horizon cannot be obtained by directly using the observation data in the past time due to time varying water demand. Water network simulator is used instead to generate the hydraulic and quality data needed for model identification. There are many uncertain factors that need to be considered. The uncertain coefficients and variables involved in the simulation problem are time varying and their variation rates are quite different, which hints that single estimation technique cannot handle all the estimation problems. Hence, different estimation techniques are needed. The paper proposes the DWDS simulator based design of chlorine dynamic model where the uncertainty is quantified by bounds. Algorithm for calculating the bounds based on the DWDS structure and demand prediction is derived.