In the centralized operation mode of thermal power plants

the abnormal vibration of steam turbines poses a serious challenge to system safety and operation stability. Focusing on the actual operation characteristics of abnormal vibration of steam turbines

the article studies the causes of vibration fluctuations and structural deviations under typical working conditions

and deeply analyzes the influence mechanism of abnormal vibration on the regulatory system. On this basis

it designs a data-driven intelligent diagnostic process to complete vibration signal collection and time-frequency feature extraction

build an abnormal recognition model based on convolutional neural network

and realize the accurate classification of multi-type vibration modes. It further formulates the control strategy linkage mechanism to improve the engineering response ability of diagnostic results

and verify the effectiveness of the strategy in combination with the actual working condition data

and quantify the inhibition effect. The research aims to provide a reliable technical path and engineering implementation basis for intelligent early warning and rapid intervention of steam turbine vibration risk in thermal power plants.