HARDWARE FAULT PREDICTION METHOD AND PREDICTION APPARATUS

摘要

Provided in the embodiments of the present application is a hardware fault prediction method, used for identifying hidden fault symptoms in hardware in advance by means of learning from historical traffic statistics data and configuration data, thereby enhancing operation and maintenance efficiency. An embodiment of the present application comprises: acquiring historical traffic statistics data and configuration data and historical warning data of an optical module; performing feature extraction on the historical traffic statistics data and configuration data to obtain historical feature data; on the basis of the historical warning data, labelling the historical feature data to obtain labelled historical feature data; performing model training on the labelled historical feature data to obtain a current classification model; and, on the basis of a preset algorithm, filtering the current classification model to obtain a reference classification model.