BP neural network-based cavitation inception judgment method for axial through-flow turbine

The invention discloses a method for judging cavitation inception of a shaft tubular turbine based on a BP (Back Propagation) neural network. The method comprises the following steps of: acquiring a time sequence x (t) after sampling radial vibration speed signals of a rotating wheel of the shaft tubular turbine in a cavitation inception state and a normal state; calculating a peak-to-peak value, a root-mean-square value and a rectification average value of the signal as radial vibration speed characteristic parameters of the rotating wheel; constructing a BP neural network model which comprises an input layer, a hidden layer and an output layer which are connected in sequence; and using the radial vibration speed characteristic parameters of the runner as a training set to train a BP neural network model, and for the trained BP neural network model, inputting a peak-to-peak value, a root-mean-square value and a rectification average value of a to-be-detected state to obtain a diagnosis result. According to t