Devising a procedure to form the diagnostic parameters for locomotives using a principal components analysis

Modern diagnostic systems are characterized by that the flow of diagnostic information requires significant computational resources to process. In order to improve the reliability of the object to be diagnosed and reduce operating costs, it is necessary to improve procedures for analyzing diagnostic results. This paper suggests a procedure to form the diagnostic features of locomotive nodes based on the use of a principal components analysis. The proposed approach is distinguished by a decrease in the dimensionality of the input set of diagnostic features in order to select the sets of interconnected diagnostic parameters. Based on the selection of the sets of interconnected diagnostic features, constructing new latent diagnostic parameters has been proposed. A latent diagnostic parameter contains information that combines data from several initial diagnostic features. The result of the method is a set of latent diagnostic parameters that do not correlate with each other and reflect the behavior of the object to be diagnosed from different technical points. The application of a sufficient number of latent diagnostic parameters involved the scree test method. This paper reports the results from using the proposed approach for treating the results from diagnosing the hydraulic transmissions in locomotives. The result from applying the procedure has made it possible to propose using three latent diagnostic parameters to assess the technical condition of a locomotive’s hydraulic transmission during bench tests. The suggested parameters contain 90 % of the original information and reflect losses in the transmission, as well as the load at the input and output of the transmission