Fault diagnosis method of parallel convolution improved triple network

The invention provides a fault diagnosis method of a parallel convolution improved triple network, which relates to the technical field of fault diagnosis, and comprises the following steps: carrying out data acquisition on operation state data of a gearbox to obtain an initial data set; constructing triple data for each piece of initial data to obtain a triple data set; constructing a fault diagnosis model which comprises a feature extraction network, a distance measurement network and a classifier network; inputting the triple data set into a feature extraction network, extracting feature vectors of the triple data set by using the feature extraction network, and inputting the feature vectors into a distance measurement network and a classifier network for joint iterative training to obtain an optimized fault diagnosis model; and acquiring real-time data of the gearbox, and performing fault diagnosis on the real-time data by using the optimized fault diagnosis model to obtain a diagnosis result. The problem