Abstract: A cigarette device management system includes an information management module, and a maintenance prediction module, a device state acquisition module and a purchase statistics module which are in communication connection with it. The maintenance prediction module is built by an artificial neural network model and is configured to predict a failure interval. The device state acquisition module acquires a device operating parameter and an operating duration of the device in real time, and compares the device operating parameter with a safety parameter range. In accordance with the system, the maintenance timing of the device can be predicted, the purchase quantity of the spare part and the purchase timing of the spare part also can be predicted.