The massive amount of data involved in the operation of numerical nuclear reactor (numerical reactor) can be used to optimize existing numerical reactor models, obtain scientific discoveries in the field of nuclear energy, and promote numerical reactor research.Based on the review of the existing data-driven modeling and the prediction of microscopic phenomena in reactors, the concept of the big data of numerical nuclear reactor was put forward, and its important characteristics as industrial and simulation big data were analyzed according to the characteristics of the field of nuclear energy.Taking China virtual reactor 1.0 (CVR 1.0) as an example, starting from the variety, dependency and inaccuracy of the big data of numerical nuclear reactor, the research work of modeling optimization and scientific discovery was carried out by using the multidisciplinary techniques such as neural network, mathematical statistics and numerical analysis, which illustrates the guiding role of the characteristics of the big data of numerical nuclear reactors in numerical reactor research.
