Abstract: To address the performance degradation of concrete caused by low-temperature curing during winter construction, a synergistic modification method combining early-age thermal activation curing and hybrid straw/basalt fiber incorporation was proposed. The mix proportions of fibers (straw fiber:2%、4%、6%; basalt fiber: 0.1%、0.15%、0.2%) were optimized, and different curing regimes were designed. Mechanical properties and frost resistance of concrete were systematically investigated. Test results indicated that the incorporation of 0.15% basalt fiber alone increased the 28-day compressive strength by 12.9%. For specimens with hybrid fiber (4% SF + 0.15% BF) under thermal activation curing, after 100 freeze-thaw cycles, the relative dynamic elastic modulus remained at 92.5%, and the mass loss ratio was only 0.8%, demonstrating frost resistance comparable to that under standard curing. A deep convolutional neural network (CNN) was further employed to establish a prediction model for freeze-thaw damage. The model achieved intelligent prediction of the dynamic elastic modulus with an R² of 0.99 and a prediction error below 3%. The results confirm that the proposed method effectively enhances the low-temperature durability of concrete, and the CNN model provides a reliable intelligent tool for evaluating frost resistance.