Locomotive optimal adhesion control by wavelet analysis

A wavelet transform method processing wheel acceleration signal containing random noise is proposed in this paper. By multiresolution analysis, the original wheel acceleration signal is decomposed into approximation parts and detail parts at several levels. After this decomposition, the noise containing in the wheel acceleration signal is stored in the detail parts. From the noise-free approximation parts, the wheel acceleration tendency can be estimated and differentiated to identify the wheel slip tendency. A fuzzy adhesion controller is then proposed to locomotive anti-slip control. The simulation results show the application of wavelet transform to the slip tendency identification paves a new way to effectively avoid false slip identification, and enhance the fully utilization of adhesive force.