چكيده به لاتين
Railway point equipment is one of the structures in railway infrastructure that is used to control the movement of trains remotely; Therefore, to increase the reliability of their operation, availability and safety of passengers, it is very important to control the health status of the components of the point machine system. The accuracy of data-based health assessment methods depends on the extraction and selection of distinguishing features of healthy and defective point machine status from raw data. The more accurately the selected features report defects in point machine performance, the more complete information will be obtained from a point machine health assessment.
In this research, the aim of this study is to investigate three feature selection methods including curve based methods (), CFS Relief and NCA on the characteristics extracted from the flow data obtained from sensors installed on the motor current of the Aperin station point machines. Therefore, more than 30 outstanding features obtained from time, frequency and time-frequency domain analysis of point machines that have healthy performance and point machines that have defects in one of the internal devices such as gear, spiral rod, slide plate, Is extracted. Then, in order to evaluate the performance in ranking among the features extracted, simple Bayesian algorithms (NB), support vector machine (SVM) are used. Feature selection methods using criteria such as detection accuracy, root mean square error, etc. Are calculated.
