Wavelet analysis of on-dimensional traction signals

The article presents issues concerning the analysis of one-dimensional traction signals describing energy conversion processes in driving systems of electric traction vehicles. The method of FFT analysis has been compared to the wavelet method. The most crucial tasks have been discussed of wavelet analyses of signals connected with the problems of drive modelling. Fundamental tasks of coefficient analyses such as approximation, compression and denoising have been taken into account, as well as more complex tasks such as filtering band component and feature extraction. An example of analysis of current signal from tram driving system obtained in the conditions of tractive adhesion failure has been presented. A hypothesis has been demonstrated that coefficients obtained in subsequent stages of wavelet decomposition can constitute the basis for indicating electric current feature vector for the phenomenon of adhesion failure. Application has been proposed of simple characteristics of transform coefficients such as energy, number of zero transitions and power. A feature vector of coefficients has been chosen. A method of feature classification has been proposed in order to implement it in adhesion failure recognition. The results of verification of wavelet analysis application in adhesion failure recognition have been presented. As tance has been taken as to the results of the diagnosis taking into consideration currently used anti-slip systems of traction vehicles.