| Resumo: | Keeping large-scale transportation infrastructure networks, such as railway net-works, operational under all conditions is one of the major challenges today. The budgetary constraints for maintenance and network operation and the network dimension are two of the main factors that make the management of a transporta-tion network such a challenging task. Hence, aiming to assist the management of a transportation network, a data-driven model is proposed for stability condition identification of rock cuttings slopes. It should be noted that one of the key points of the proposed system is to avoid data from complex monitoring equipment or laboratory expensive testes. Accordingly, only information taken from routine in-spections (visual information) and complemented with geometric and geologic data will be used to feed the models. Therefore, in this work the flexible learning capabilities of Artificial Neural Networks (ANN) were used to fit a data-driven model for Earthwork Hazard Category (EHC) identification. Considering the high number of parameters involved in EHC identification, Genetic Algorithms (GA) were applied for input feature selection purposes. The proposed models were addressed following a nominal classification strategy. In addition, to over-come the problem of imbalanced data (since typically good conditions are much common than bad ones), three training sampling approaches were explored: no resampling, SMOTE and Oversampling. The achieved modelling results are pre-sented and discussed, detailing GA effectiveness and ANNs performance.
|
|---|