Abstract: A comprehensive fault diagnosis method was proposed for lubrication state of journal bearings based on empirical mode decomposition (EMD) and weighted grey relational degree, since a mapping relation exists between the energy distribution in different frequency bands of acoustic emission and the lubrication state. First, the acoustic emission signals were decomposed into a finite number of stationary intrinsic mode functions based on EMD algorithm, then false components contained in the IMF was eliminated using the correlation coefficient method, and finally the first 10-order IMF components containing main fault information were chosen to calculate the energy ratio and to construct the characteristic vector. Results show that the weighted grey relational analysis has good classification effect on recognition of small samples, which can be used to calculate the grey incidence of different acoustic emission signals, so as to perform fault diagnosis on lubrication state of journal bearings effectively.