Unforeseen data estimation in water distribution system

Abstract

Unforeseen data is a mostly encountered problem in the water distribution system (WDS) application. The WDS is considered to be significant component in the social framework. This paper proposes an improved version of window based Kalman filter known to be customized Kalman filter (CKF). The academic section WDS of National Institute of Technology-Tiruchirappalli, Tamil Nadu, India is considered as the case-study. The flows across the pipes and head level of the reservoir tanks are measurement of interest which has significant data loss due to the randomly varying sampling interval. The available measurements are also corrupted with the state and input dependent noises along with the conventional uncertainties. This is due to the interconnection between the consumers and structural complexity of the considered case-study which causes the direct and indirect impact over the other states due to the propagation of the disturbance occurring at any spatial instance of the interconnected WDS. The proposed CKF is equipped with the necessary remedies to have an optimal estimates of unforeseen data even in the presence of state and input dependent noises. The CKF algorithm is incorporated along with the mean imputation technique as a priori of the estimates, which improves the estimation accuracy. The estimated results are validated through the quantative and qualitative analysis for the considered case-study.