GIS service life prediction method based on electrical, chemical, structural and vibration parameters

The invention relates to a GIS service life prediction method based on electrical, chemical, structural and vibration parameters, and the method comprises the steps: collecting a large amount of sample data during the normal operation and fault of GIS equipment, forming a data set, and dividing the data set into a training set and a test set; establishing an FA-BP neural network of the BP network optimized by the FA, inputting the training set into the FA-BP neural network, and outputting a weak predictor model Gt (x); integrating all weak predictor models Gt (x) by using an AdaBoost algorithm, and establishing a strong predictor; and obtaining a qualified life attenuation function model G (x), inputting the test set into the qualified life attenuation function model G (x), and outputting a predicted life value. The problems that a traditional prediction method cannot accurately and comprehensively reflect the actual service life of the GIS equipment, the prediction precision is low, and preventive maintenanc