Optimal sensor location for overland flow network monitoring
Résumé
The problem of optimal sensor location for monitoring of an overland flow network is addressed in this paper. The flow dynamics in each branch is described by the continuity equation of Saint-Venant model and Green-Ampt infiltration formula, while the initial states are assumed to be unknown. A methodology is then proposed to optimize the placement of sensors, with the purpose of estimating those states by a least square error minimization between measured and estimated water heights. Considering the method of variational calculus to solve for this minimization, a criterion based on the computed adjoint state is used to maximize the output sensitivity by appropriate sensor locations. A case-study of three small surface runoffs connected together at a point where two upper flows stream down into the third one is considered as an application example, and numerical simulations are provided with different measurement scenarios to validate the proposed optimal sensor location technique as well as the estimation method.