A Random Matrix Theoretical Approach to Early Event Detection Using Experimental Data
In this paper, High-dimensional data analysis methods are proposed to deal with random matrix which is composed by the real data from power network before and after the fault. The mean spectral radius (MSR) of non-Hermitian random matrices is defined as a statistic analytic for the fault detection. By analyzing the characteristics of random matrices and observing the changes of the spectral radius of random matrices, grid failure detection will be achieved. This paper describes the basic mathematical theory of this big data method, and the real-world data of a certain China power grid is used to verify the methods.