Sound Source Localization-Based Gas Leakage Detection in Railway Pneumatic Brake Systems

The sealing effect of train braking systems is directly related to their operational safety and reliability. Gas leakage will cause the system braking effect to decline and lead to brake failure if the degree is severe. Micro-leakage due to pipeline joints or valve bodies is caused by the vibration, temperature fluctuations, and fatigue occurring after prolonged use. The traditional manual detection method is a time-consuming procedure that does not suit the high-speed railway fast maintenance demands. Based on the GCC-PHAT source localization algorithm, a handheld detection method for train braking pipelines is proposed. A 7-channel circular microphone array continuously records the leakage sound, while spectral analysis, adaptive band-pass filtering, and GCC-PHAT extract the frequency components containing characteristics of fault information. The effective frequency is located precisely and visualized. It validates the system via experiments carried out on the simulated braking pipeline platform, and proves the validity and portability of the prototype and scheme. Results show the system produces stable acoustic energy map within sound pressure level of 20-80dB with localization errors less than 5cm, and provides high quality maintenance mode for train braking system.