Aiming at the scarcity of spectrum resources in smart grid, the integration technology of cognitive radio and smart grid is used to improve the utilization of spectrum resources in smart grid. To solve the problem of optimal resource allocation in cognitive smart grid, a chaotic swarm elite leader and Riemannian manifold salp swarm algorithm is designed. Halon sequence chaos is used to initialize the salp population, which enhances the diversity and enables the algorithm to quickly lock the optimal solution range. In order to avoid leaders falling into local optimization, swarm elites are used to randomly replace the position updates of leaders, so as to avoid leaders falling into local optimization and improve the search ability of leaders. In addition, a mutation strategy combining Riemannian manifold and student t-distribution is proposed to enhance the activity of the population and overcome the defect that the algorithm falls into local optimization due to population aggregation in the later stage. The IEEE CEC benchmark function set is used to test the effectiveness of the improved algorithm, and the curves are drawn for effectiveness analysis. In order to apply modified chaotic adaptive salp swarm algorithm(MCASSA) algorithm to the application potential of resource allocation solution in cognitive smart grid, a comparative analysis is carried out with the goal of maximizing the transmission rate of smart grid, and the fairness and maximum benefit of users are compared and analyzed. Experimental results show that MCASSA algorithm can effectively improve the performance and resource utilization of cognitive smart grid.