A neural network approach for railway safety prediction

Artificial neural networks (ANNs) are becoming increasingly popular for solving complex problems, as they can behave quite well at solving problems that don't have an algorithmic solution or for which the algorithmic solution is too complex to be found. In railway systems, the problem of predicting the system malfunctions, or equivalently, railway safety is of paramount interest for most of railway companies. Traditional ways of predicting railway safety are very expensive in terms of time consuming, which make them inefficient under certain circumstances. This paper advocates the use of ANNs architecture to handle the safety problem. By taking irregularities in the positioning of the rails as input to the ANN, the ANN can predict the safety ratio of the rails. In order to reduce the dimensionality of inputs data a wavelet transformation technique has been employed. Different neural network structures are created and their performances both in terms of mean squared error and correlation coefficient have been evaluated to find out the best structure for predicting railway safety. The experiments showed that when the model is trained on a dataset subset and then tested on different subset, it performed satisfactorily and can predict the desired output with a very low error factor.