Aplicação da análise de séries temporais para detecção e prognóstico de danos em estruturas inteligentes

• Autor(a) principal: Cano, Wagner Francisco Rezende [UNESP]
• Data de Publicação: 2015
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://hdl.handle.net/11449/127896; http://www.athena.biblioteca.unesp.br/exlibris/bd/cathedra/02-09-2015/000847631.pdf
• topic: Monitoramento da integridade estrutural; Localização de falhas (Engenharia); Lamb, Ondas de; Structural health monitoring

This work presents an approach based on time series processing to deal with the damage detection and prognosis issue in structures coupled with piezoelectric sensors and actuators considering eventual operational and environmental variabilities. The first approach is based on the identification of a predictive autoregressive model obtained with a reference time response. Damage indicative metrics are extracted from prediction errors and the separation of effects (loading or damage) is performed by a fuzzy clustering algorithm. This procedure is carried on a composite structure attached to a material test system to reproduce loading conditions in order to simulate real operational conditions. On the other hand, the second proposed methodology employs a two step identification. First, an autoregressive model is created for structural monitoring similarly to the previous procedure, but employing statistical process control to detect progressive damage. Next, autoregressive models with exogenous inputs are estimated for reference and damaged conditions in order to track variation of parameters, allowing the prognosis of the structure's future structural condition. Initial tests on an aluminum plate indicated that this method is capable of performing a reasonable prognosis and predicting structure's dynamic behavior associated to a specific level of mass reduction. Both methods and results are discussed and compared by the end of the work