Method for discrimination of time series data value for analysing vehicle gear vibration

摘要

The method involves first embedding the time series data value y(t) in an n-dimensional reconstructed phase space and then selecting an arbitrary data vector Xi from the trajectories of the of the embedded time series data value. In the form of a Euclidian distance m data vectors Xj (j=1,2,...m) are selected which are adjacent to the selected data vector and the tangential unit vectors Ti and Tj relative to the data vectors Xi and Xj are derived. Deviations in the directions of the adjacent data vectors relative to the tangential unit vectors Tj are calculated drawing on the following expression for the unit vectors. This calculation for gamma i is repeated for a specific sample number K, which corresponds to sub-spaces, in order to derive a mean value GAMMA . Following this it has to be determined whether GAMMA is not equal to 0. From the result it can be determined whether the time series data value is based on a causal or random signal