Optimal sensor placement for contamination detection: A multi-objective and probabilistic approach

Water networks are spatially dispersed, easily accessible, and vulnerable to contaminating intrusions. If contamination is detected too late, then damage to the population may be irreversible. For the hard task of optimal sensor placement, this work presents a multi-objective approach that is combined with post-processing methods for a Pareto front analysis. The contamination is represented by the chemical reactions of the pesticide Parathion in water quality simulations. A multi-objective approach is used that incorporates four contamination probability functions. The Pareto front is analyzed with a clustering approach, and a coverage matrix is used to evaluate the centers of each cluster. An automatic selection solution method, based on distances to the most suitable solution, is also explored.