The invention relates to a hydroelectric generating set vibration
trend prediction method. The method comprises the steps of obtaining a sample
data set of hydroelectric generating set vibration
monitoring data, analyzing the vibration
monitoring data, drawing a relation coordinate graph of a to-be-predicted vibration parameter or throw parameter and other related parameters, and determining candidate characteristic parameters; a CMI-SAL prediction model is established; and screening unit key equipment vibration related working condition parameters through a condition
mutual information method module. The method has the beneficial effects that the input characteristics are screened by applying a
conditional mutual information correlation analysis method, two or more vibration variables can be analyzed, the relevance and correlation among the variables can be judged, the over-redundancy defect is overcome on the basis of a
mutual information method, and the prediction efficiency is improved; by applying a sliding window and a maximum
pooling method and adopting a convolutional layer, the
order of magnitude scale of input data can be reduced, a local maximum value is extracted from input features, the number of trainable parameters is reduced, and the data robustness and the operation speed of the model are improved.