Ammeter fault diagnosis method and device of stacked recurrent neural network

The invention discloses an ammeter fault diagnosis method and device of a stacked recurrent neural network. The method comprises the following steps: obtaining current time sequence data associated with a to-be-diagnosed ammeter, inputting the preprocessed current time sequence data into a trained stacked recurrent neural network model, and carrying out the fault diagnosis of the to-be-diagnosed ammeter to obtain the output Y^ corresponding to the current time sequence data, wherein the Y^ is an L-dimensional vector, and L is equal to the total number of fault categories needing to be judged by the model; and determining the diagnosis category of the to-be-diagnosed ammeter according to the output corresponding to the current time sequence data. According to the invention, different types of ammeter faults can be effectively represented, and automatic diagnosis of the ammeter faults is realized; moreover, the input of the stacked recurrent neural network model is current time sequence data, and dynamic time s