Aiming at the problems of low solution accuracy, slow convergence speed, and susceptibility to local optima in the rime optimization algorithm (RIME), a multi-strategy fusion-based RIME algorithm has been proposed. Firstly, population initialization was done using Tent chaotic mapping to enhance initial population diversity. Secondly, an enhanced inverse learning mechanism was employed to generate populations with better fitness values, thereby improving the algorithm's convergence speed. Finally, a differential evolution strategy was introduced to increase population diversity, preventing the algorithm from falling into local optima and improving both population utilization and solution accuracy. Numerical experiments were conducted using the CEC 2017 test function, and the results show that the improved algorithm outperforms other comparative algorithms. Lastly, the improved algorithm was applied to feature selection problems, with 10 datasets from the UCI database selected for simulation experiments. The results indicate that the improved algorithm effectively enhances data classification accuracy.