Control performance improvement using unknown disturbance estimation based on kalman filter

When controlling a stochastic LTI system under white Gaussian system noise and measurement noise with a PID controller, it is difficult to obtain a good steady-state response due to the malfunction of the D control action. This problem is typically solved using a separation principle based on the Kalman filter to reduce the influence of noises. However, not even the PID control system based on the Kalman estimator can follow the reference input in the steady state when unknown step disturbance is applied, so a steady state error is inevitable. This paper suggests solving this problem, using a method to comprise a fuzzy disturbance estimator. The estimated disturbance is fed back to the input part of the controlled system to reduce the influence of the unknown disturbance and is also fed back to the Kalman filter to compensate the filtered estimate; the control performance of the suggested overall control system is proven through several simulations. © ICROS 2018.