In order to develop the theory and method of grid situation awareness based on big data technology, inspired from random matrix theory, this paper proposes a dynamic identification method for abnormal state detection in low SNR Environment based on spiked population model. It firstly constructs a data source matrix and obtains a moving split-window matrix and its standard matrix, then acquiring the sample covariance matrix. The global SNR estimation is performed using the classical spectral estimation method corrected by the Kaiser window function, from which a corresponding dynamic threshold is derived. In this way, the situation awareness and early warning for interconnected power systems could be achieved by calculating maximum eigenvalue of sample covariance matrix and the dynamic threshold to check violation. Utilizing MATLAB? software, the case studies have been carried on an IEEE 50-machine system, involving two main working conditions such as abnormalSload change and short circuit fault, the results shows that the proposed methodology has the advantage of higher noise resistance in comparison with the traditional mean spectral radius based method and preliminarily verifies that it wouldSbeSrobustSunderSincompleteSinformation.