A wagon train health system, based on the acceleration measurements on a wagon car body, has been designed to monitor the wagon safety performance indexes - derailment ratio, car body or bogie hunting, speed, maximum dynamic wheel load and wheel unloading due to long-wavelength track geometry irregularities. In this paper, the possiblity to monitor the wheel impacts due to short-wavelength defects such as wheel flats using such a system is theoretically presented through the simulations using a comprehensive non-linear vehicle-track interaction dynamics model. The method is restricted to tradtional three piece bogie rollingstock. A coal wagon is modeled and a section of track with the geometry irregularity class 5 is selected for the simulations. The simulation results show that the original acceleration measurements on the wagon car body include very high frequency components contributed by the friction elements in the wagon components, which hide the useful messages coming from the wheel impact. A filter has been designed to extract these useful messages. The processed data shows that the wheel impact patterns, even those caused by the smaller wheel flats, can be detected. As expected, wheel impact detection using this approach is limited at lower speeds.