This　paper　presents　a　novel　control　strategy　to　improve　the　performance　of　a　low-frequency　vibration　calibration　system.　The　low-frequency　vibration　calibration　system　based　on　a　linear　motor　is　first　developed.　The　amplitude　sensitivity　and　phase　sensitivity　are　then　derived.　Subsequently,　the　composite　control　strategy　of　the　iterative　learning　control　(ILC)　and　Luenberger　observer　is　established　to　improve　the　performance　of　the　linear　motor　vibration　generator.　Finally,　the　frequency　stability　of　the　linear　motor　is　measured,　and　a　tri-axial　accelerometer　is　calibrated.　Experimental　results　indicate　that　the　linear　motor　controlled　by　the　proposed　composite　strategy　can　fulfill　accelerometer　calibration　with　high　precision,　and　maintain　the　performance　in　the　full　frequency　band.