Filtering a Random Loading Process in the Problem of Assessing the Durability of Rolling Stock Parts

The article addresses the urgent problem of selecting a filtering method for a random loading process, which in some cases may be non-stationary. This issue is closely related to the problem of durability assessment, as its solution determines how completely the oscillation components (including high-frequency ones) are represented in the output load vector. The problem is analysed using loading processes obtained from operational measurements of a real object at several commonly used movement speeds. Several filtering methods are analysed, including the use of low-frequency median filters, the level method, and modal decomposition. Since the processing involves subsequent durability assessment, conditional durability evaluations were performed to demonstrate the similarities and differences between the filtering methods in the calculations for the filtered processes. Recommendations for the preliminary processing of the data are provided.