Estimating the Concentration of Nitrates in Water Samples Using PSO and VNS Approaches

摘要

In this paper we present a study of the application of a Particle Swarm Optimization (PSO) and a Variable Neighborhood Search (VNS) algorithms to the estimation of the concentration of nitrates in water. Our study starts from the definition a model for the Ultra-violet spectrophotometry transmittance curves of water samples with nitrate content. This model consists in a mixture of polynomial, Fermi and Gaussian functions. Then, optimization algorithms must be used to obtain the optimal parameters of the model which minimize the distance between the modeled transmittance curves and a measured curve (curve fitting process [1]). This process allows us to separate the modeled transmittance curve in several components, one of them associated to the nitrate concentration, which can be used to estimate such concentration. We test our proposal in several laboratory samples consisting in water with different nitrate content, and then in three real samples measured in different locations around Madrid, Spain. In these last set of samples, different contaminant can be found, and the problem is therefore harder. The PSO and VNS algorithms tested show good performance in determining the nitrate concentration of water samples.