Aiming at the problems of imperfect standards for minor repairs of test equipment and unsound specifications for major and medium repairs, this paper proposes a particle swarm-support vector machine-based decision-making model for major, medium and minor repairs of test equipment on the basis of combing historical maintenance data. Based on the preprocessing of historical maintenance data, the particle swarm inertia factor is optimized by linearly decreasing weights to improve the global search and local convergence capabilities of the particle swarm algorithm, and the global optimization capabilities of the particle swarm algorithm are used to optimize the penalty coefficient and kernel of the support vector machine. The function width improves the classification accuracy and accuracy of the model. The model optimized through sample data training is used for the classification and decision-making of the equipment maintenance level, and is compared with the classification results of the BP neural network and