Abstract: Massively uses the frequency Shift Keying Signal（FSK） in our country locomotive control signal to transmit each kind of locomotive speed information.Therefore，fast accurately detection the FSK signal to have the extremely important practical significance to the locomotive fast safe operation.The wavelet analysis is one extremely effective method，that analyzes the non-steady signal and Mallat fast algorithm makes the practical application of wavelet analysis possible.This article has first studied the locomotive FSK signal characteristic，proposed a kind of FSK signal real-time processing method based on Mallat and FFT，and the method has been simulated and realized on TMS320C5409 signal processing system.The testing results are given in the paper，and indicates that this method is one effective handling for processing real-time locomotive signal.

Key words: wavelet transformation, Mallat algorithm, FFT, locomotive signal, DSP simulation

摘要： 在我国机车控制信号中大量采用移频键控信号（FSK）来传送各种机车速度控制信息。因此，快速准确地检测FSK信号对机车快速安全运行有着十分重要的现实意义。小波分析是分析非稳定信号的一种非常有效的方法，而Mallat快速小波算法使得小波分析的实际应用成为可能。首先研究了机车FSK信号的特征，提出了基于Mallat和FFT相结合的FSK信号实时处理方法，该方法在TMS320C5409组成的信号处理系统上仿真实现。给出了实际采集信号的处理结果，结果表明该方法是实时处理机车信号的一种有效方法。

关键词: 小波变换, Mallat算法, 快速傅立叶变换, 机车信号, DSP仿真