Vibration sensors are involved extensively in a variety of applications. Especially in the era of the Internet of Things, developing self-powered vibration sensors has become a very meaningful yet challenging problem. This study investi- gates a highly sensitive self-powered vibration sensor based on the triboelectric nanogenerator (VS-TENG) for machinery condition monitoring. By introducing a stacked structure comprising foamed aluminum, and a fluorinated ethylene propylene film with gold-plated electrode protected by two indium tin oxid layers, The VS-TENG can detect the vibrations with frequencies ranging from 1 to 2000 Hz and can detect vibrations of low amplitude (approximately sub-μm, by calculation). The output performance and characteristics of the TENG under various vibration frequencies, accelerations, and amplitudes are analyzed system- atically. The VS-TENG is successfully used to monitor the operating conditions of mechanical gear systems, reaching a recognition accuracy of 99.78%. Further- more, it can also be utilized for vibration detection in other areas such as the air compressor, heat gun, hollow tile recognition, etc., which means a solid progress toward the practical applications of TENGs in self-powered vibration detection.