With the increasing complexity of the power grid and the integration of large-scale wind power, the uncertainty of power grid continues to grow. Accurate judgment of the self-organized critical state of the power system has important application value for the prevention and control of cascading failures. To tackle this challenge, from a data-driven perspective, a new self-organized critical state identification algorithm for power systems with wind power integrated based on random matrix spectrum analysis theory is proposed. Firstly, the initial sample random matrix is established by using fluctuating line load rate data. Secondly, the self-organized critical state discriminant matrix is constructed on the basis of the stable initial sample random matrix. Then, the matrix linear eigenvalue statistics and the inner ring radius in the single ring theorem are used to construct the self-organized critical state discrimination index and related criteria. The simulation on the IEEE-39 bus system shows that the proposed discrimination index can track the changes of the verification index more effectively than other methods. And it can give the threshold for the system to enter the critical state of self-organization.