REAL-TIME FORECASTING FOR WATER LEVELS IN SEWER BY MACHINE LEARNING AND INUNDATION SIMULATION

Abstract

 This paper presents a practical forecasting method for water level in sewer by using a machine learning. A physical model is usually used to calculate water levels. However, it takes a lot of time to calculate them. This is a serious issue for sewer management in real-time in an emergency. In this paper, this issue is solved by using the machine learning. In addition, we proposed hybrid forecasting which also learns the results of physical simulation because the machine learning cannot predict unprecedented case. As a result, the calculation time and accuracy in proposed forecasting are improved.