Remote　and　on-line　measurement　of　chromium　on　structural　steel　surface　in　nuclear　power　plants　is　critical　for　protection　against　fluid　accelerated　corrosion.　To　improve　the　insufficient　sensitivity　of　fiber-optic　laser-induced　breakdown　spectroscopy　toward　trace　element　detection,　a　dual-pulse　spectral　enhancement　system　is　set　up.　In　an　iron　matrix,　for　the　purpose　of　improving　sensitivity　of　trace　chromium　analysis　and　reducing　the　self-absorption　of　iron,　the　effects　of　key　parameters　are　investigated.　The　optimal　values　of　the　parameters　are　found　to　be:　450　ns　inter-pulse　delay,　700　ns　gate　delay,　30　mJ/6　mJ　pulse　energy　ratio,　and　19.8　mm　lens-to-sample　distance　(corresponding　to　a　799　μm　laser　focused　spot　size).　Compared　to　the　single-pulse　system,　the　shot　number　of　dual-pulse　ablation　is　limited　for　reducing　surface　damage.　After　the　optimization　of　the　dual-pulse　system,　the　signal-to-noise　ratio　of　the　trace　chromium　emission　line　has　been　improved　by　3.5　times　in　comparison　with　the　single-pulse　system,　and　the　self-absorption　coefficient　of　matrix　iron　has　been　significantly　reduced　with　self-reversal　eliminated.　The　number　of　detectable　lines　for　trace　elements　has　more　than　doubled　thus　increasing　the　input　for　spectral　calibration　without　significantly　increasing　the　ablation　mass.　Three　calibration　methods　including　internal　standardization,　partial　least　squares　regression　and　random　forest　regression　are　employed　to　determine　the　chromium　and　manganese　concentrations　in　standard　samples　of　low　alloy　steel,　and　the　limit　of　detection　is　respectively　calculated　as　36　and　515　ppm.　The　leave-one-out　cross　validation　method　is　utilized　to　evaluate　the　accuracy　of　chromium　quantification,　and　the　concentration　mapping　of　chromium　is　performed　on　the　surface　of　a　steel　sample　(16MND5)　with　a　relative　error　of　0.02　wt%.　©　2020