The invention relates to a
corrosion fatigue life prediction method based on a BP neural network and application. The prediction method comprises the following steps: selecting maximum stress,
stress ratio, loading frequency and pH value of a solution as main factors influencing
corrosion fatigue life; designing and
processing a
corrosion solution circulating device matched with a
corrosion fatigue test, and carrying out a
corrosion fatigue circulation failure series experiments on a high-strength
sucker rod sample in a specific production environment, collecting and neatening experiment data and dividing the experiment data into training samples and prediction samples; setting
artificial neuron network parameters, and establishing nonlinear mapping between the influencing factors and the
corrosion fatigue life; training and testing a
nervous system; and predicting the corrosion fatigue life of a new sample. The corrosion fatigue life prediction method based on the BP neural network has the beneficial effects that the corrosion fatigue life of a high-strength
sucker rod is predicted by high non-
linear approximation capability of the BP neural
network model, and operation is simple; and the prediction method is high in generalization performance, and
engineering application is facilitated.