In　this　paper,　a　prediction-correction　scheme　is　proposed　to　online　measure　and　regulate　the　milling　depth　of　microchannel　using　an　indicator　of　laser　triggered　plasma.　Firstly,　a　prediction　model,　with　respect　to　the　laser　fluence　and　feedrate,　is　established　with　several　calibration　tests　using　the　least　square　fitting　method.　It　is　utilized　to　change　the　focal　position　of　objective　to　track　the　depth　evolution　of　newly　generated　surface.　Meanwhile,　a　scanning　path　for　every　milling　layer　with　an　offset　in　Z-axis　at　the　beginning　and　the　end　of　the　trajectory,　is　developed　to　drive　the　plasma　brightness　periodically　changing.　Then,　the　milling　depth　could　be　obtained　when　the　brightness　reaches　to　the　maximum　value.　By　doing　so,　an　online　measurement　method　is　presented　to　estimate　the　milling　depth　using　the　trend　of　plasma　brightness.　Furthermore,　a　correction　model　is　developed　to　iteratively　adjust　the　feedrate　with　the　online　estimated　depth.　Therefore,　the　microchannel　milling　process　could　be　monitored　and　controlled　in　a　closed-loop　manner,　in　order　to　accurately　regulate　the　milling　depth.　Finally,　an　online　measurement　and　closed-loop　microchannel　milling　is　carried　out　on　the　self-developed　micro-machining　center.　The　effectiveness　and　correctness　of　the　proposed　method　are　verified　by　comparing　the　estimated　depth　with　the　actually　measured　results.　©　2016　Elsevier　Ltd