The selfsensing magnetic bearing could reduce the cost and the axial size of the magnetic bearing and increase its reliability. A selfsensing control method based on mixedkernel least squares support vector machine (LSSVM) forecasting model was proposed for an axial active magnetic bearing (AAMB). The principle and mathematical model of the active magnetic bearing were introduced; based on the principle of the mixedkernel LSSVM, the nonlinear forecasting model between the current and the displacement which realized the displacement selfsensing control was built through parameter optimization. The control system of the AAMB with selfsensing was constructed. The simulation results showed that the prediction model could accurately detect the rotor axial displacement. Further experimental results also showed that the method had a good ability of axial displacement selfsensing. The AAMB could realize stable suspension operation without displacement sensors.