Prognosis and fail detection in a dynamic rotor using artificial immunological system

Author(s)

Outa, Roberto ^[1] ; Chavarette, Fabio Roberto ^[2] ; Mishra, Vishnu Narayan ^[3] ; Goncalves, Aparecido C. ^[4] ; Roefero, Luiz G. P. ^[4] ; Moro, Thiago C. ^[4]

Affiliation

^[1] FATEC Aracatuba, Fac Technol Aracatuba, Dept Biofuels, Aracatuba - Brazil

^[2] Univ Estadual Paulista, Fac Engn Ilha Solteira, Dept Math, Ilha Solteira - Brazil

^[3] Indira Gandhi Natl Tribal Univ, Dept Math, Amarkantak - India

^[4] Univ Estadual Paulista, Fac Engn Ilha Solteira, Dept Mech Engn, Ilha Solteira - Brazil

• Document type: Journal article
• Source: ENGINEERING COMPUTATIONS; v. 37, n. 9 APR 2020.
• Web of Science Citations: 1

Abstract

Purpose In recent years, the mechanical industries began to apply many investments in research and technological development to obtain efficient methods to analyze the integrity of structures and prevent disasters and/or accidents, ensuring people's lives and preventing economic losses. Any structure, whether mechanical or aeronautical, before being put into use undergoes a structural integrity assessment and testing. In this case, non-destructive evaluations are performed, aiming to estimate the degree of safety and reliability of the structure. For this, there are techniques traditionally used such as ultrasonic inspection, X-ray, acoustic emission tests, among other techniques. The traditional techniques may even have a good instrumental apparatus and be well formulated for structural integrity assessment; however, these techniques cannot meet growing industrial needs, even more so when structures are in motion. The purpose of this paper is to demonstrate artificial immune systems (AISs), ate and strengthen the emergence of an innovative technological tool, the biological immune systems and AISs, and these are presented as computing methods in the field of structural health monitoring (SHM). Design/methodology/approach The concept of SHM is based on a fault detection mechanism used in industries, and in other applications, involving the observation of a structure or a mechanical system. This observation occurs through the dynamic response of periodic measurements, later related to the statistical analysis, determining the integrity of the system. This study aims to develop a methodology that identifies and classifies a signal in normal signals or in faults, using an algorithm based on artificial immunological systems, being the negative selection algorithm, and later, this algorithm classifies the failures in probabilities of failure and degree of fault severity. The results demonstrate that the proposed SHM is efficient and robust for prognosis and failure detection. Findings The present study aims to develop different fast access methodologies for the prognosis and detection of failures, classifying and judging the types of failures based on AISs. The authors declare that the present study was neither published in any other vehicle of scientific information nor is under consideration for publication in another scientific journal, and that this paper strictly followed the ethical procedures of research and publication as requested. Originality/value This study is original by the fact that conventional structural integrity monitoring methods need improvements, which intelligent computing techniques can satisfy. Intelligent techniques are tools inspired by natural and/or biological processes and belong to the field of computational intelligence. They present good results in problems of pattern recognition and diagnosis and thus can be adapted to solve problems of monitoring and identifying structural failures in mechanical and aeronautical engineering. Thus, the proposal of this study demonstrates and strengthens the emergence of an innovative technological tool, the biological immune system and the AIS, and these are presented as computation methods in the field of SHM in rotating systems - a topic not yet addressed in the literature. (AU)

• FAPESP's process: 19/10515-4 - Prognosis and mechanical structure failure detection using natural computing
• Grantee: Fábio Roberto Chavarette
• Support Opportunities: Regular Research Grants