A recurrent neural network for wastewater treatment plant effuents' prediction

• Ivan Pisa ^[1] ; Ignacio Santín ^[1] ; Jose Lopez Vicario ^[1] ; Antoni Morell ^[1] ; Ramon Vilanova ^[1]
  1. [1] Universitat Autònoma de Barcelona

     Universitat Autònoma de Barcelona

     Barcelona, España

     
• Localización: XXXIX Jornadas de Automática: actas. Badajoz, 5-7 de septiembre de 2018 / coord. por Inés Tejado Balsera, Emiliano Pérez Hernández, Antonio José Calderón Godoy, Isaías González Pérez, María Pilar Merchán García, Jesús Salvador Lozano Rogado, Santiago Salamanca Miño, Blas Manuel Vinagre Jara, 2018, ISBN 978-84-9749-756-5, págs. 621-628
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Wastewater Treatment Plants (WWTP) are industries devoted to process water coming from cities' sewer systems and to reduce their contamination. High-pollutant products are generated in the pollutant reduction processes. For this reason, certain limits are established and violations of them are translated into high economic punishments and environmental problems. In this paper data driven methods are performed to monitor the WWTP behaviour. The aim is to predict its effluent concentrations in order to reduce possible violations and their derived costs. To do so, an alarm generation system based on the application of Artificial Neural Networks (ANNs) is proposed. The proposed system shows a good prediction accuracy (errors around 5%) and a reduced miss-detection probability (30%).