The invention relates to a method for identifying a health degradation state of a rolling bearing. The method comprises the steps of: (1), obtaining a historical monitoring signal of the rolling bearing, pre-processing the monitoring signal, eliminating a singular value, and reducing noise; (2), respectively performing time-domain analysis, power spectrum analysis and CEEMDAN decomposition on thepre-processed monitoring signal, so that time-domain, power spectrum and intrinsic mode energy characteristics are obtained; (3), training a CSVM model by utilizing various characteristics of the monitoring signal; and (4), for the online real-time acquired monitoring signal of the rolling bearing, inputting the various characteristics obtained in the step (2) into the CSVM model trained in the step (3), so that the current health degradation state identification result of the rolling bearing is obtained. According to the method in the invention, the health degradation state of the rolling bearing can be identified precisely in real time; real-time monitoring on the state of the rolling bearing is realized; and thus, safe, steady and long-period operation of a numerical control machine tool is ensured.