Based　on　orthogonal　array　testing　strategy　(OATS),　the　effects　of　sand-binder　ratio　(S/B),　water-binder　ratio　(W/B),　and　the　ratio　of　steel　fiber　volume　to　reactive　powder　concrete　(RPC)　volume　(STF/R)　on　the　compressive　strength　and　chloride　diffusion　coefficient　of　RPC　were　investigated　using　an　artificial　neural　network　method.　Research　results　reveal　that　the　compressive　strength　of　RPC　approaches　summit　when　STF/R　is　2%　or　W/B　is　0.18-0.2%,　and　decreases　with　the　increasing　of　S/B.　Furthermore,　the　chloride　diffusion　coefficient　increases　with　W/B　or　STF/R　and　decreases　with　S/B.