Abstract: To replace operator's patrol inspection, an early warning system for safety interlock malfunction was designed for ZB45 packing line by adopting artificial intelligence technology. Via rebuilding the data acquisition system and extracting the characteristic information related to protective covers, the characteristic data set for predicting the effectiveness of safety interlocking was constructed, and the prediction model was established based on non-linear support vector machine algorithm. The prediction model for the safety interlock of outlet protective cover in a cartoner was tested as an example, and the results showed that the prediction precision of the model reached 99.2% with the accuracy rate of 94.7% and the recall ratio of 94.7%, the precision increased by 3.1 percentage points and the recall ratio raised by 47.4 percentage points comparing with manual inspection. This technology provides a support for promoting the safety management level of cigarette production equipment.