Abstract: With the continuous development of power technology,the scale and complexity of power grid are constantly increasing.It results in corresponding power data increasing exponentially.The conventional model method can not make full use of historical and real-time data,and has inevitable assumptions and simplifications.However,the random matrix theory method driven by data does not rely on specific mechanism modeling.This paper firstly describes large dimensional statistical analysis and analyzes the important role of random matrix principle with consideration of the limitations of classical multivariate statistical analysis.Then the concepts of single ring theorem and linear spectral statistics are emphatically introduced.The highdimensional statistical index is derived,i.e.,the mean spectral radius.Combined with the existing knowledge system,the application basis and data sources of the random matrix theory are analyzed,and the concrete idea of power grid state analysis is given.Thereby the disturbance positioning method of power grid based on voltage data is proposed.Finally,the effectiveness of the method is verified through several examples.