Aiming at the problems of low quality

information isolation and distortion of power equipment operationand maintenance data

a two-layer knowledge base architecture is proposed. A mathematical model for the knowl-edge generation of operation and maintenance and defect rate of equipment under different characteristic combina-tions are constructed by using model unification

block statistics and dynamic sorting algorithm

with the correlationcoefficient between defect rate and factors as the knowledge representation. Through the continuous accumulation ofoperation and maintenance samples

defect rate of equipment under different characteristic combinations and thecorrelation between the defect rate and factors in the operation and maintenance knowledge base is updated

so as toimprove the accuracy of defect diagnosis. Based on the defect data of power company operation and maintenance

thevalidity of the knowledge generation method based on two-layer architecture are analyzed and verified

which canprovide theoretical basis for equipment operation and maintenance decision.