The environment of offshore platform is complicated and its safety problem is becoming more and more serious. It is necessary to model and analyze the structure of offshore jacket platform by finite element method. Due to structural corrosion damage and other problems occurring during the service period of the platform, there will be significant differences between the finite element model built according to the design drawings and the real structure. Therefore, the finite element model should be revised according to the actual situation. In this paper, a finite element model correction method based on random forest is proposed, which can give the modelcorrection parameters quickly and accurately. Firstly, different design parameters of the platform were selected, then parameters to be modified were obtained through design, finally, corresponding fundamental frequencies were obtained ,respectively. Pearson correlation analysis and the fundamental frequency to get the parameters with large correlation coefficient with the fundamental frequency. The experimental parameters were redesigned to obtain the fundamental frequency. Random forest network was used to reduce the nonlinear difference between the two to verify the accuracy and generalization ability of the network were. Finally, combined with the measured data, the effect of the finite element model modification is compared and analyzed. The results show that the regression fraction of the model can reach 0.998, and the modified finite element model can more accurately reflect the dynamic characteristics of the real platform.