The phenomenon of locomotive wheel tread irregularities seriously affects the safety of vehicle operation. This paper introduce the constitute and working principle of a locomotive wheel tread detection system.In order to improve the efficiency of irregularity’s recognition, this paper presents an algorithm based on wavelet analysis and modulus maxima in the MATLAB environment. This algorithm use wavelet to analyze the data and get the approximate and detail coefficients of data signal, than we use the modulus maxima method to analyze the singularity of data and orientate the position of data point where have an tread irregularity further. The results of experiment indicate that the effect of this locomotive wheel tread irregularities processing algorithm is obviously, and it can accurately judge the tread irregularities.