Abstract: This study employed the Support Vector Machine (SVM) method combined with the Particle Swarm Optimization (PSO) algorithm to optimize the penalty coefficient C and the radial basis function width coefficient g of the SVM model for predicting the oxide layer thickness of TiBw/Ti55 composites based on oxidation experiments. The optimized penalty coefficient (C) is 64.15405, and the radial basis function width coefficient (g) is 0.56689. A predictive model for the oxidation layer thickness of TiBw/Ti55 composites was established. The established model for predicting the oxide layer thickness of TiBw/Ti55 composites is f\left(x\right)= \displaystyle\sum\nolimits_i,j=1^16\left(\alpha _i-\alpha _i^*\right) \exp\left(-0.56689\left|\right|x_norm^i-x_norm^j\left|\right|^2\right)+ 0.39490 . The predicted model has a training set coefficient of determination (R²) of 0.98706 and a root-mean-square error (RMSE) of 0.05445. The testing set has a coefficient of determination (R²) of 0.99504 and a root-mean-square error (RMSE) of 0.13211. The average error between the predicted and experimental values of the oxide layer thickness of TiBw/Ti55 composites is 5.88%, indicating that the SVM model established in this study has high predictive accuracy. On the basis of the above, the influence of oxidation temperature, volume fraction of the reinforcing phase, and oxidation time on the oxide layer thickness of TiBw/Ti55 composites is analyzed.