Failure prognosis in a piston flow reactor of the tubular reactor using artificial immune systems

Abstract

Industry 4.0 uses principles of new technologies to achieve better quality and productivity results, which intelligent computing is one of them. In this work is proposed an intelligent methodology in which the artificial immune system sequesters the signal into groups and determines the classification based on failure prognose by the degree of severity of a tubular reactor with piston flow. The process is developed as follows: basically, after obtaining the vibration signals from the reactor through a numerical model, the Fourier rapid transform is used to transform the signals in the frequency domain.  Later, a negatively select artificial immune system performs the diagnosis, identifying and classifying failures. The motivation for the application of this methodology is the process of supervision of structures, in order to identify and characterize failures, as well as make decisions aimed at avoiding accidents or disasters. The results demonstrate the accuracy and the robustness of the methodology for the reactor operation.