| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 12.75 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A presente dissertação foca-se no desenvolvimento e aplicação de uma metodologia de deteção e identificação de danos nas rodas de um veículo ferroviário de mercadorias do tipo Laagrss, baseada em respostas dinâmicas induzidas por este na via. Foi adotado o sistema de monitorização Wayside composto por um conjunto de acelerómetros instalados nos carris da via-férrea, em que a identificação do dano nas rodas estabelece-se por um processo indireto. Foi dado especial enfase à deteção e identificação de anomalias relacionadas com imperfeições na circularidade das rodas, denominadas de poligonização. O conjunto de dados representativo das condições das rodas foi adquirido com o recurso a modelos numéricos de interação entre o veículo e a via considerando diferentes cenários. Assente num método não supervisionado e remoto, foi aplicado neste trabalho uma metodologia de monitorização da condição das rodas do veículo ferroviário, com recurso a técnicas de análise e tratamento de dados multivariados baseadas em inteligência artificial. Foi realizada a extração de indicadores sensíveis ao efeito da poligonização nas rodas com recurso a modelos autorregressivos (AR) e autorregressivos com entradas exógenas (ARX), análises de componentes principais (PCA) e transformadas wavelet (CWT). Posteriormente, foi utilizado técnicas de normalização de dados em relação a fatores ambientais e operacionais (baseados em PCA). Finalmente, foi desenvolvido e aplicado técnicas de classificação de dados capazes de distinguir estados com e sem dano baseadas em análises Outliers e análise de Clusters para identificação da severidade do dano. A metodologia prova ser eficaz na deteção do dano com resultados bastante satisfatórios, relativamente à identificação da severidade são verificadas algumas falhas.
This dissertation focuses on the development and application of a methodology for wheel damage detection and identification of a Laagrss type rail freight vehicle, based on dynamic responses induced by it on the track. The Wayside monitoring system was adopted, consisting of a set of accelerometers installed on the track rails, where wheel damage identification is established by an indirect process. Special emphasis was given to the detection and identification of anomalies related to imperfections in the circularity of the wheels, called polygonisation. The data set representative of the wheel conditions was acquired with numerical models of interaction between the vehicle and the track considering different scenarios. Based on an unsupervised and remote method, a methodology for monitoring the condition of the railway vehicle wheels was applied in this work, using multivariate data analysis and processing techniques based on artificial intelligence. The extraction of features sensitive to the effect of wheel polygonization was performed using autoregressive (AR) and autoregressive models with exogenous inputs (ARX), principal component analysis (PCA) and wavelet transforms (CWT). Subsequently, data normalisation techniques were used in relation to environmental and operational factors (based on PCA). Finally, data classification techniques capable of distinguishing states with and without damage based on Outlier analysis and Cluster analysis were developed and applied to identify damage severity. The methodology proves to be effective in detecting the damage with very satisfactory results. Regarding the identification of the severity, some flaws are verified.
This dissertation focuses on the development and application of a methodology for wheel damage detection and identification of a Laagrss type rail freight vehicle, based on dynamic responses induced by it on the track. The Wayside monitoring system was adopted, consisting of a set of accelerometers installed on the track rails, where wheel damage identification is established by an indirect process. Special emphasis was given to the detection and identification of anomalies related to imperfections in the circularity of the wheels, called polygonisation. The data set representative of the wheel conditions was acquired with numerical models of interaction between the vehicle and the track considering different scenarios. Based on an unsupervised and remote method, a methodology for monitoring the condition of the railway vehicle wheels was applied in this work, using multivariate data analysis and processing techniques based on artificial intelligence. The extraction of features sensitive to the effect of wheel polygonization was performed using autoregressive (AR) and autoregressive models with exogenous inputs (ARX), principal component analysis (PCA) and wavelet transforms (CWT). Subsequently, data normalisation techniques were used in relation to environmental and operational factors (based on PCA). Finally, data classification techniques capable of distinguishing states with and without damage based on Outlier analysis and Cluster analysis were developed and applied to identify damage severity. The methodology proves to be effective in detecting the damage with very satisfactory results. Regarding the identification of the severity, some flaws are verified.
Description
Keywords
Deteção e Identificação de danos Veículo ferroviário Laagrss Respostas dinâmicas Sensores Sistema de monitorização Wayside Poligonização nas rodas Réplicas digitais numéricas Tratamento de dados multivariados Inteligência artificial Extração de indicadores Modelos autorregressivos (AR) Modelos autorregressivos com entradas exógenas (ARX); Análises de componentes principais (PCA) Transformadas wavelet (CWT) Análises Outliers Análise de Clusters Damage detection and identification Railway vehicle Dynamic responses Sensors Wayside monitoring system Wheel polygonization Numerical digital replicas Multivariate data processing Artificial intelligence Features extraction Autoregressive models (AR) Autoregressive models with exogenous inputs (ARX) Principal component analysis (PCA) Wavelet transforms (CWT) Outlier analysis Cluster analysis