Detection of incipient failures by using an H2-norm criterion: application to railway switching points

The paper deals with condition monitoring of electric switch machines for railway points. The proposed detection system is based on off-line processing of armature current and voltage sampled as the machine operates a switch of railway tracks. These data are normally available from conventional railway signalling infrastructures; no additional transducers and instrumentation are required. The system, which basically consists of an algorithm tuned on a model of the machine behaviour as both the rails are simultaneously driven towards their rest position, allows detection of faults of an incremental nature, specifically those connected to progressive increasing of frictional loads due to loss of lubrication, deterioration of slide chairs and increasing obstructions. The algorithm implements finite impulse response systems whose convolution profiles are designed on the basis of an H2-norm type criterion which guarantees robustness, particularly with respect to electrical noise.