Abstract Summary
This work focuses on the development and application of a methodology for detecting wheel polygonization in a Laagrss type rail freight vehicle based on dynamic responses induced by it on the track. To achieve that, a numerical train-track interaction model was adopted to simulate the passage of a train over a wayside monitoring system composed by a set of accelerometers installed on the rails. Based on an unsupervised 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 normalization 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 and Cluster analysis were developed and applied to identify damage severity. The methodology proved to be effective in detecting the damage with very satisfactory results. Regarding the identification of the severity, some flaws are still verified.
