The invention discloses a factory intelligent early-
warning system which comprises the following successively connected modules: a multivariate heterogeneous data fusion
processing module which collects
monitoring data generated by each
system of a factory and performs fusion on
mass multivariate heterogeneous data based on
xml technology, and performs
processing through statistics methods such as sampling and normalization; a
machine learning module which performs training learning on data that are generated by the multivariate heterogeneous data fusion
processing module and performs continuous adjustment for generating a
neural network system model which is related with fault data and fault phenomena and is used for analyzing the multivariate heterogeneous data that are generated in a subsequent production
system and performing forecasting; and an information early-warning module which intelligently pushes an analysis result that is generated by the
machine learning module to related management personnel for reminding. According to the factory intelligent early-
warning system, through directly using a
deep learning neural network system which is established according to
deep learning technology in analytical forecasting for real-time
mass data of the factory, high efficiency and high intelligence are realized, and furthermore factory management by the people is facilitated in an automatic evolution manner.