The　magnetizing　ability　of　excitation　structureshas　great　influence　on　the　detectability　of　magnetic　flux　leakage　sensor.According　to　the　method　of　AC　magnetic　leakage　detection,　a　parametric　finite　element　model　of　magnetic　flux　leakage　detection　is　established.　The　influences　of　the　shape　and　size　of　the　magnetic　yoke,　the　position　of　the　exciting　coil,　the　winding　length　and　the　thickness　of　the　magnetic　shielding　layer　on　the　detection　signal　are　investigated.　In　addition,　an　optimization　method　of　the　size　of　the　excitation　structure　is　developed　by　combining　parametric　finite　element　analysis　with　genetic　algorithm.The　distance　between　the　magnetic　poles　and　the　width　of　the　magnetic　pole　is　optimized.　The　simulation　and　test　results　show　that　the　excitation　structure　has　a　great　influence　on　the　results　of　magnetic　flux　leakage　testing,　and　the　optimized　magnetic　excitation　structure　is　helpful　to　improve　the　detection　performance　of　the　magnetic　flux　leakage　sensor.　The　proposed　optimization　method　provides　a　feasible　method　for　improving　the　performance　of　magnetic　flux　leakage　sensors.　©　2018　Journal　of　Mechanical　Engineering.