Abstract: To address the limitations of traditional natural frequency adjustment methods for the rotor blades of blast furnace top gas recovery turbine unit (TRT), such as narrow adjustment ranges of frequencies and the tendency to change the aerodynamic performance of the rotor blades, an automatic natural frequency adjustment method based on the response surface model and the multi-objective genetic algorithm was established by introducing optimization principle. This method transforms the natural frequency design problem into an optimization problem, and enables the automatic adjustment of the natural frequency of the TRT rotor blades, which consists of five major steps: First, parameterize each three-dimensional rotor blade by fitting the two-dimensional blade profiles of different cross-sections along spanwise with Bezier curves to achieve flexible control of the blade profile, taking the vertical coordinates of the control points of these Bezier curves as the design variables of the optimization problem. Second, obtain samples with these design variables at different levels through the Latin hypercube experimental design method, and conduct modal analysis on each sample based on the finite element method to obtain the first four-order natural frequencies of each sample. Third, take the natural frequencies that needs to be adjusted according to the design requirements as the target natural frequencies, and establish a response surface model between the design variables and the target natural frequencies based on the Kriging model and the experimental sample data. Fourth, use the multi-objective genetic algorithm in combination with the established response surface model to perform optimization, obtain the optimal design variables, and figure out the optimized blades. Finally, conduct a flow field simulation on the optimized blades to obtain the aerodynamic performance and determine whether the optimization result is feasible. Two computational cases were conducted with the rotor blades of a two-stage TRT. To verify the efficacy of the automatic frequency adjustment method, Case 1 implemented traditional frequency adjustment techniques on the rotor blades and the resultant frequencies were adopted as benchmarks, which means the objective function is set to minimize the difference between the blade's natural frequencies obtained by the automatic frequency adjustment method and these frequencies. Case 2 employed the frequency avoidance rate criterion to define the target natural frequency spectrum, to validate the ability of the automatic frequency adjustment method to improve the natural frequency. Results show that when achieving the same magnitude of frequency adjustment, under the design working conditions, both the flow rate and efficiency of the traditional frequency-adjusted blades decrease significantly compared with those of the original blades, while the flow rate and efficiency of the automatic-adjusted blades are almost the same as those of the original blades; within the design space, the automatic frequency adjustment method can obtain blades that meet the design requirements of the natural frequency by more flexibly controlling the distribution of the blade profile; generally, thickening the blades increases the natural frequency of the blades, while the opposite leads to a decrease in the natural frequency. This research work has reference value for the development of high-efficiency and high-reliability TRT blade design technology.