To diagnose the faults occurred in pulverizing system accurately and rapidly

this paper proposed a novel method for constructing the process memory matrix

against the pulverizing system fault diagnosis method based on nonlinear state estimation

according to the operation characteristics and fault features of the coal milling system.This new method is used to pretreat fault data based on Mahalanobis distance.In addition

it establishes process memory matrix based on the pretreated fault data to reduce redundancy data.By this way

the new method improves the fault diagnosis efficiency and plays a more active role in ensuring practicability and stability of the coal milling system fault diagnosis.