Abstract: Cement-based structural health monitoring materials have important application value in the field of civil engineering. At present, cement-based self-sensing composites generally face problems such as the difficulty in balancing electrical conductivity and mechanical properties, as well as insufficient sensitivity for damage monitoring. To address the above challenges, this study innovatively proposes a cement-based self-sensing material system synergistically reinforced by graphene conductive coatings and copper-coated steel fibers, and prepares specimens with different proportions through screen printing technology. The results of three-point bending loading tests show that under the optimal proportion (steel fiber content of 2.5% and graphene coating thickness of 0.5 mm), the resistance change rate of the material reaches 181% during the main crack propagation stage, and the peak flexural strength is 3.57 kN, demonstrating excellent force-electricity collaborative response capability. By using the main crack length and maximum deflection, a dimensionless damage index \varPsi is constructed. This index realizes continuous quantification of structural damage states and safety grading by coupling the main crack length and maximum deflection, providing a clear basis for damage assessment in engineering applications. Additionally, a Multilayer Perceptron (MLP) machine learning regression model is introduced to achieve high-precision nonlinear mapping between resistance and the damage index, with the determination coefficient R² of the model's test set reaching 0.961. The research results show that this composite system effectively improves the sensitivity and accuracy of damage monitoring, providing a new technical path for the intelligent assessment of the health status of cement-based structures.