DEVELOPMENT OF A NEW ALGORITHM FOR WHEEL-RAIL CONTACT FORCE ESTIMATION USING REDUCED MEASUREMENT SETS

In the present work, the problem of the contact forces measure is analysed with a particular focus on the dynamometric wheelset method. Knowing the state of deformation on the wheelset, it is possible to reconstruct the forces that caused it after a calibration process. In this research, the definition of the minimal measurements sets required for a consistent estimation of the forces is defined, highlighting the need of six independent measurement sections for the estimations of the vertical and lateral forces. It is important to point out that dynamometric wheelset are very delicate systems and may suffer of damages or drifts on some measurement sections, which can affect or compromise the possibility to estimate correctly the wheel-rail contact forces. It is indeed important to derive a method for the measurement of the forces also in case of degraded conditions of the dynamometric wheelset in order to increase the reliability of the measuring system. By means of a proper partitioning of the calibration test-plan, it is possible to achieve a series of calibration matrices, each one suitable for a specific calibration conditions. These calibration matrices can provide good results in terms of estimation precision even though the measurement set is reduced below the minimal conditions. At this point, a method based on the fuzzy logic is derived in order to recognise properly the conditions encountered by the dynamometric wheelset. This method is capable to identify between the calibration models the most suitable. Eventually, the algorithm is also capable to mix the models in order to cover the regions out of the calibration ranges. The effectiveness of the method is tested by means of quasi-static tests performed on a full-scale dedicated test-rig highlighting the possibility to increase the performances of a non-optimally instrumented wheelset in terms of both forces reconstruction and measurement precision.