Batch processes data is three dimensional data composed of lots of the batches

variables and time. The data contains abundant useful global and local structure information for process monitoring. The key of fault diagnosis is to fully extract the feature information of batch process. The traditional methods unfold three-dimensional data to two-dimensional data. The process inevitably leads to destructing internal structure of the data. The traditional methods usually only consider the global information data or local information data

then the useful process information can not be fully extracted

which leads to poor diagnosis. Aiming to above problems

a tensor global-local neighborhood preserving embedding (TGNPE) algorithm is proposed in this paper. First tensor factorization is used to deal with three-dimensional data directly which effectively save the internal structure of the data. Then the neighborhood preserving embedding algorithm is used to extract the local structure of the data information

at the same time considering the global information of the data. The data information can be fully extracted under keeping internal data structure. The contribution plot method is used to diagnose fault variables after detecting faults. The simulation results of penicillin fermentation process verified the effectiveness of the proposed algorithm.