The power big data has the characteristics of large scale

fast generation and dynamic. The difficulty in de-fining the neighborhood relationship between the data results in the increase of computation. Therefore

an incremen-tal attribute reduction method of power big data based on neighborhood relationship matrix is proposed

in which theneighborhood relation set is transformed into neighborhood relation matrix

and the single attribute neighborhood rela-tion matrix is optimized by combining the sorting idea and the characteristics of neighborhood relation matrix. On thisbasis

matrix heuristic information is set up. Then

the neighborhood relationship matrix of each attribute is calculat-ed

condition attributes are added one by one and the attribute correlation degree is calculated

attributes are select-ed for the reduction subset according to their degrees of attribute importance

the and the whole process is iterateduntil the importance degree does not change

thus incremental attribute reduction of power big data is realize. Experi-mental results show that the present method needs shorter computation time of attribute reduction

and has better re-duction effect and higher reliability.