Unmanned　Aerial　Vehicles　(UAVs)　are　playing　an　increasingly　critical　role　in　military　surveillance　missions.　However,　due　to　safety　and　economic　issues,　it　is　necessary　to　validate　the　UAV　performance　and　algorithms　in　a　semi-physical　simulation　environment　before　the　real　flights.　In　this　study,　the　mechanical　structure　and　control　system　of　the　semi-physical　simulation　system　is　developed　for　the　UAVs.　In　order　to　better　simulate　the　attitude　of　the　UAV　during　a　flight,　the　flight　attitude　simulator　is　designed.　Also,　a　sand　table　of　the　mountainous　scenery　from　real　inspection　tasks　is　developed　to　work　with　the　simulation　system.　Path　planning　algorithms　are　embedded　in　the　platform,　and　improvements　and　evaluations　of　traditional　algorithms　are　carried　out.　Then,　an　improved　adaptive　particle　swarm　optimization　(IAPSO)　algorithm　is　proposed　to　improve　the　accuracy　of　the　path　of　UAVs.　An　energy-consumption　prediction　model　is　established　for　the　platform.　By　combining　the　IAPSO　algorithm　with　the　energy-consumption　prediction　model,　the　best　path　with　low　energy　consumption　can　be　obtained.　Finally,　the　effectiveness　of　the　simulation　system　is　verified　by　comparison　experiments　of　the　mathematical　simulation,　semi-physical　simulation,　and　real　flight.