Abstract: A fault time determination based on the normal distribution and a fault identification based on the eigenvector method are proposed. The data source matrix is constructed by using the three-phase voltage data which is collected by the PMU of each node. The fault time is determined through analysing the normal distribution curve of each column data in the time window matrix. Then, the judgment matrix of eigenvector method is constructed with the sample covariance matrix. The fault type is identified by analysing the anomalous element of eigenvectors of the two largest eigenvalues. The IEEE 39 nodes system is used to verify that the method can effectively achieve the fault time determination and the fault identification when all kinds of faults and complex faults occur in the power grid. And it is also proved that the method has a certain anti-interference ability to the bad data.