Abstract: Active chemical powder has important application value in repairing concrete micro-cracks,improving mechanical strength,enhancing durability,prolonging service life and reducing maintenance cost. In order to obtain excellent performance of self-healing concrete,this study uses complexing agent SMS,auxiliary agent (aluminum silicate + sodium sulfate,with a mass ratio of 1 ∶1),calcium formate and sodium gluconate as concrete admixtures for orthogonal experiment. The self-healing properties of concrete are evaluated by key parameters such as compressive strength and splitting tensile strength,and the effect law of active chemical powder on the self-healing properties of concrete is studied. Experimental studies have shown that different proportions of active chemical powder and different degrees of damage have an important influence on the strength of concrete. For the multi-factor complex experimental results,the study uses the method of machine learning to build a self-healing concrete strength recovery performance prediction model,takes the initial strength and damage repair strength as optimization objectives,and combines the good and bad solution distance method to propose the mix ratio optimization method of active chemical components,so as to obtain the mix ratio of active powder with the best strength recovery performance of self-healing concrete. Experimental verification and mechanism explanation have been carried out. This study is helpful to deeply understand the effect of active chemical powder on the self-healing properties of concrete and guide the design of construction mix ratio. 

Key words: active chemical powder,compressive and splitting strengths,strength recovery,mix ratio optimization,healing mechanism