In the stabilization of inverted pendulum, the angle is polluted by the measurement noise. When making use of
the angle signal to calculate the angular velocity through difference, the noise is amplified, which makes the controller output
oscillate randomly. Besides, the motor, as the actuator, is polluted by the plant noise, leads to the strong system oscillation,
which not only is harmful for pendulum stabilization, but also reduces the motor’s service life. Combining the steady-state
Kalman filter with the disappearing memory method, the effective way to determine the Kalman filter gain matrix is found
out, which overcomes the model error, satisfies the real time requirement, decreases the motor’s oscillation, and gets the
better performance.