Honing mesh is crucial for maintaining the lubrication state between piston and cylinder liner, and shallowed mesh caused by wear will cause lubrication deterioration and intensified friction and wear, and even major mechanical accidents such as cylinder pulling. An acoustic emission-based method for detecting the wear depth of honing mesh on cylinder liner was put forward, which identified and extracted the feature variables related to the depth of mesh from the acoustic emission signals, and established the mapping relationship between the acoustic emission feature signals and the depth of mesh by constructing a convolutional neural network model. Based on a longstroke reciprocating friction tester, acoustic emission signals of honing mesh specimens with different depths were collected and analyzed under variable rotational speeds and loads, and it was found that there was a significant correlation between the depth of honing mesh and the response characteristics of the acoustic emission signals in the low-frequency and highfrequency regions. Based on the proposed model, the detection and evaluation of earlier wear failure on mesh was carried out. The results showed that the average accuracy of honing mesh depth recognition reached 94%, which verified the feasibility of non-intrusive detection of mesh wear depth based on acoustic emission.