This study constructed a unified framework of “experimental characterization deep learning” and systematically tested the multi age mechanical properties, water transport durability, and long-term shrinkage characteristics of incineration biomedical waste ash (IBMWA) concrete with a dosage of 0-12.
5%; Combining SEM image quantitative segmentation to extract microstructure and morphology parameters, a microstructure index (MI) was proposed to characterize the equilibrium between gel phase and defect phase.
The study determined 7.
5% as the optimal cement replacement dosage for IBMWA, corresponding to a 90 day compressive strength of 57 MPa; The deep learning model trained based on SEM microscopic features achieved a prediction accuracy of 95.
1%, R2=0.
92， A new paradigm for strength prediction driven by microstructure has been established.
.