The invention discloses an intelligent electric energy meter fault classification detection method and system based on a self-coding network. The method comprises the steps of performing normalizationprocessing on acquired historical detection data of an intelligent electric energy meter, and dividing the historical detection data into a training set and a test set; initializing and setting parameters of the self-coding network model; selecting sample data from the training set, inputting the sample data into the self-coding network model to obtain signal characteristics, inputting the signalcharacteristics into a classifier for classification, and performing iterative training according to a classification result; continuously adjusting the parameters of the self-coding network model according to the classification result of the test set so as to determine the optimal parameters of the self-coding network model; and carrying out classified detection on faults of the intelligent electric energy meter by using the self-coding network model corresponding to the optimal parameter. According to the method, the collected signals are subjected to unsupervised feature extraction throughthe deep noise reduction self-coding network, fault signals can be rapidly and accurately classified, the fault recognition capability of the intelligent electric energy meter can be improved, and compared with a traditional method, the method has high noise resistance.