The invention discloses a
water turbine vibration fault diagnosis method based on time-
frequency analysis and a residual network, and relates to the technical field of fault diagnosis, and the method comprises the steps: collecting and preprocessing the vibration data of a
water turbine, obtaining the preprocessed vibration data, and constructing a vibration fault
data set; performing model training by using the vibration fault
data set, and constructing a vibration fault diagnosis model; and performing fault diagnosis based on the vibration fault diagnosis model to obtain a vibration fault diagnosis result. According to the invention, unified
processing of vibration data with different lengths is realized, the problem of inconsistent vibration data lengths in actual
engineering is effectively solved, and the
utilization rate and representativeness of the data are improved; the
convolution operation is performed in combination with batch normalization and an
activation function, so that the gradient disappearance problem in
deep learning is effectively relieved, and data features can be learned more fully; accurate recognition of no fault,
single fault and multiple faults is achieved, the misjudgment rate is reduced, and the credibility of the diagnosis result is improved.