为了研究玄武岩纤维在盐类侵蚀环境下改善再生混凝土的孔隙性能，通过电镜扫描技术分析混凝土内部变化规律，并建立混凝土孔隙率与干湿循环次数、渗透高度、抗压强度以及劈裂抗拉强度之间的线性回归关系模型。对不同浸泡龄期和纤维掺量下的渗透高度建立全连接神经网络（fully connected neural network, FCNN）模型，用于预测玄武岩纤维再生混凝土在盐类侵蚀环境下的抗侵蚀寿命。结果表明：孔隙率随着侵蚀龄期的增加逐渐增加，适量掺入玄武岩纤维能显著降低混凝土孔隙率，其中纤维掺量为1.0%时改善效果最为显著；随着纤维掺量的增加，再生混凝土抗盐渗透能力增加，即渗透高度降低；当纤维掺量为1.0%时，可获得最佳的抗压和劈裂抗拉强度；FCNN模型结果表现良好，为玄武岩纤维再生混凝土的寿命预测提供了可靠的参考。

In order to study the effect of basalt fiber on the porosity of recycled concrete under salt erosion, the internal variation of concrete was analyzed by scanning electron microscopy, and a linear regression model was established between the porosity of concrete and the number of dry and wet cycles, the height of penetration, the compressive strength and the splitting tensile strength. A fully connected neural network (FCNN) model was established for the penetration height under different soaking ages and fiber content for predicting the corrosion resistance life of basalt fiber regenerated concrete under salt erosion conditions. The results show that the porosity of concrete increases gradually with the erosion age increasing, and the appropriate addition of basalt fiber can significantly reduce the porosity of concrete, and the improvement effect is most significant when the fiber content is 1.0%. With fiber content increasing, the salt permeability resistance of recycled concrete increases, and permeability height decreases. When the fiber content is 1.0%, the best compressive strength and splitting tensile strength are obtained. The fully connected neural network model has a good effect and provides a reliable reference for the life prediction of basalt fiber recycled concrete.