The invention discloses a real-time analysis method for correlation between the low-frequency vibration of a steam turboset and the temperature of lubricating oil in the fields of the vibration state monitoring and the fault diagnosis of rotary machinery. The method comprises the following steps of: collecting axial relative vibration data, the rotating speed signal and the bonded phase signal ofa rotor and the numerical value of the lubricating oil temperature of a bearing; calculating and storing low-frequency vibration amplitude entropy at the current time at every two step lengths t; storing the numerical value of the lubricating oil temperature of the bearing; calculating the incremental trend parameter of the lubricating oil temperature of a rotor bearing and the descending trend parameter of the low-frequency vibration amplitude entropy when reaching the set duration; and finally judging the correlation between the low-frequency vibration and the lubricating oil temperature. In the invention, the recognition accuracy of the correlation between the low-frequency vibration and the lubricating oil temperature is improved to provide guarantee for the safe operation of the steam turboset.