The wear monitoring of metal-polymer plain bearing (MPPB) remains a critical challenge in mechanical systems. With the rapid development of triboelectric nanogenerators (TENG), the self-powered sensing technology with the triboelectric effect exhibits great potential in bearing monitoring. This study pioneers a self-powered triboelectric sensing approach for the abrasive wear detection of MPPBs, utilizing a diamond-like carbon (DLC)-coated steel shaft to generate triboelectric signals without structural modification of the MPPB bushing. The system enables abrasive wear detection based on the triboelectric effect under both dry and boundary lubrication conditions, and can balance the mechanical durability simultaneously. Different wear situations are simulated to explore the influence of the wear area and wear depth on the output signals. Furthermore, real MPPBs with different wear states have been used to validate the simulation results under both dry and boundary lubrication conditions. Distinctive signal patterns emerge corresponding to specific wear conditions, and their underlying mechanisms are elucidated systematically. Finally, a deep learning neural network is employed to accurately identify wear states based on the triboelectric signals, which can achieve a remarkably accurate rate. This research demonstrates a triboelectric-based approach for effective wear condition monitoring in MPPBs, offering potential for the development of smart bearings.