Development of an AI-based Digital Twin model for wastewater treatment plant

The process of digitalization has become a must in recent years. This digitalization affects not only to data management and communication, but also to most of the processes of the industry. One of the paradigms that have recently gained most attention is the creation of digital twins, i.e., accurate models of processes or elements that can be used to simulate their expected behavior under several possible conditions and scenarios. This modeling makes possible to reach a higher level of optimization thank to the enriched information that it provides. In the case of wastewater treatment plants, the creation of models of each part of the process could be used to identify disparities between the real values measured in the plant and those expected values provided by the models. These disparities are usually related with problems or degradation in the plant elements, unexpected events, or other contingencies, so the study of their values allows to implement predictive maintenance strategies. As part of the GEDIAV-H20 project, several monitored variables from a wastewater treatment plant were modeled using artificial intelligence techniques. It can be observed in the case study that the models can effectively predict the expected behavior of the processes in the plant.