The research behind Process Chemometrics Lab @ GEPSI –CIEPQPF



Integrating the network structure topology in process monitoring

As a result of our past research efforts, we have identified a central aspect of industrial systems that has not been properly handled so far, even considering the state-of-the-art data-driven analysis methodologies, and especially in what concerns to industrial process systems: the explicit incorporation of systems network topology along with information about the causal directionality between connected elements, in data-driven analysis methods. With such descriptions available, more precise and coherent with the true nature of systems, improvements could also be achieved in the subsequent process engineering tasks, such as systems monitoring, optimization and control. Therefore, we are developing a new data-driven analysis framework to handle data collected from complex systems, which is focused on inferring its connective and causal structure, while being able to handle several of its known multivariate and multiscale specificities: the M2Net.