A fuzzy comprehensive assessment method of running condition was constructed and applied to a large-scale centrifugal compressor set in a petrochemical corporation aiming at the monitoring and early warning of abnormal conditions in industry. The maximal information coefficient (MIC) correlation analysis of indexes was introduced to determine the independent indexes to be assessed

and the dynamic deterioration degree was calculated using the predicted independent indexes by the second-order Markov chain model. The fuzzy membership degree weighting method was employed to assess the running condition of all units in the set. Simple and direct radar chart was used to visualize condition assessment grades. Results showed that the proposed fuzzy comprehensive assessment method successfully assessed the running condition of the set. The constructed method achieved 10 min earlier alarm than the traditional threshold alarm occurred at the specific moment of 00:44 on April 7 of 2018. The method would provide a valuable tool and have a wide engineering application in ensuring the safety and reliability of industrial production.