In order to improve the output quality of railway vehicle signal communication system and reduce the output error code, it is necessary to design the signal noise reduction. A signal denoising filtering algorithm based on time-frequency characteristic decomposition and matched filtering is proposed for railway vehicle signal communication system. Firstly, the signal detection model of the railway vehicle signal communication system is constructed, and the transmission signal of the railway vehicle signal communication system is sampled. The time-frequency characteristic decomposition method is used to realize the conversion of the railway vehicle noise signal from the time domain to the frequency domain. The characteristic value of envelope amplitude of railway vehicle noise signal is extracted. According to the envelope amplitude response of the transmission signal of railway vehicle signal communication system, the autocorrelation matching filtering detection processing is carried out. The feedback modulation method is used to design the channel equalization for the railway vehicle signal communication system. The time-frequency characteristic decomposition and matched filtering are used to realize the filtering detection and de-noising filtering of the railway vehicle noise signal. The simulation results show that this method can improve the balanced transmission performance of the railway vehicle noise signal, and the output error code of the railway vehicle signal communication system is lower than that of the railway vehicle signal communication system, which can improve the balanced transmission performance of the signal, and the anti-interference performance is better. The transmission quality of railway vehicle signal is improved.