Predicción de contaminantes usando sensores inteligentesaplicación práctica en la Bahía de Algeciras

  1. Francisco Javier González Enrique
Supervised by:
  1. Ignacio José Turias Domínguez Director

Defence university: Universidad de Cádiz

Year of defence: 2021

  1. Juan Manuel Górriz Sáez Chair
  2. José Manuel Jerez Aragonés Secretary
  3. David Alberto Elizondo Giménez Committee member

Type: Thesis

Teseo: 671575 DIALNET


The Bay of Algeciras is considered one of the main industrial areas in Spain, housing very relevant petrochemical and metallurgical industries. In addition, Algeciras is one of the most prominent European ports, with significant container traffic. It is also a highly-populated area, with important cities such as Algeciras, to reach a total population of close to 300,000 inhabitants. Air pollution is one of the most important problems affecting people's quality of life, especially in highly industrialized and densely populated areas. Its control is essential to assure good air quality, thereby avoiding the possible negative impacts of pollution on the population. In this sense, atmospheric monitoring is a powerful tool to identify and evaluate problems in air quality, acting as a preventive measure aimed at safeguarding the health of the population. In addition, it provides very useful data for decision-making in relation to the environment. Pollutant forecasting models show a growing boom in recent years since they constitute a tool that allows defining the actions to be taken and the control strategies to follow to effectively treat those episodes in which the concentration of a certain pollutant may exceed the limits established in the applicable regulations. In the field of air quality networks, in recent years, there has been a growing interest in the possible advantages that could be obtained by merging the measurements obtained with different sensors. With this new approach, instead of using a single sensor, information from different sensors would be used to produce a better measurement or better prediction. This would make it possible to have various measures that would ensure the reliability of the data, avoiding errors, improving the response time and helping to improve the interpretation of the observed phenomenon. The main objective of this thesis is focused on developing models aimed at predicting the concentration levels of several different pollutants in the Bay of Algeciras. Pollution data are provided by a network of sensors located in different strategic points of the Bay (made up of different monitoring stations). In addition, it is intended to build models that can provide these sensors with intelligent characteristics, such as self-calibration capabilities, self-taring, or missing values, for example, for which a multi-sensor fusion approach will be applied using at each point the environmental information that is considered most relevant.