4-Weighted ordinary least square algorithm for batch processing.

Abstract

A deconvolution algorithm for a linear system with finite memory is developed using exponential weighting. The algorithm is developed with the idea of using it in batch mode. The solution derived is similar in nature to the ordinary least squares solution of the deconvolution problem for linear systems. The limitation of the existing procedures whereby analysis is based on a policy of equal weighting for all measured data as the process evolves is highlighted. It is suggested that this algorithm may be more suitable in total response modelling type of analysis than in a unit hydrograph type of study where events are treated in isolation.