Monitoramento da integridade estrutural de um rotor utilizando sistemas imunológicos artificiais

• Autor(a) principal: Merizio, Igor Feliciani
• Data de Publicação: 2021
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: por
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://hdl.handle.net/11449/205079
• topic: Structural integrity monitoring; Artificial immune systems; Negative selection algorithm; Fault detection; Mechanical systems; Monitoramento de integridade estrutural; Sistemas imunológicos artificiais; Algoritmo de seleção negativa; Detecção de falhas; Sistemas mecânicos

In this research work, the results obtained by the Artificial Immune System (AIS) based on the Negative Selection Algorithm (NSA) optimized by the ClonalG Selection Algorithm are displayed and discussed. NSA is based on the observation of the functioning of the human immune system, more specifically on the negative selection of T-type cells that occurs in the thymus. The operating process is divided into two phases: Censor phase and Monitoring. In the first stage, the AIS is trained with the signals collected from the structure in normal operation, this is called structure’s own sign and stored as a base-line of the structure. So, the AIS can learn about the functioning of the system and be able to identify failures. From a signal in a structural failure situation (obtained by executing the experiment until a longer time than the project), the developed AIS discriminates the signals as own and non-own (non-failure and failure) after verifying the affinities of these with the base-line. While ClonalG is based on the immune system’s continuous learning capacity, being able to learn by reinforcement and memory: When it encounters an antigen that is similar to one previously presented; this being the mechanism of vaccines, for example. Two signal banks were used, collected in different periods with different sizes of the same structure. The AIS was developed for the structure and afterwards optimization by ClonalG was applied, comparing the results obtained. The location of the fault was also performed for each case.