7. Kalman Filter - JoshSilver8/ENGR454_Buck_Converter GitHub Wiki

The Kalman Filter is similar to the full order observer in many ways. It can be thought of as the optimal full-state estimator given knowledge about measurement noise and disturbances. Based on the magnitudes of the measurement noise and disturbances, it decides whether to rely more on the measurements or on the model. If there are large disturbances, it will rely more on the measurements. Whereas if there is more noise, then it will rely more on the model. A crash course of the theory behind the Kalman filter by Steve Brunton is found here.

Again we are implementing the same Block Diagram with an optimized gain matrix K. to do this We used the LQE function in MATLAB to find the optimized gain values for the estimator. The LQE function takes into account the disturbance and noise of the system to develop this gain matrix. The results are largely the same as that found in the full order observer. Disclaimer: As we were lacking on time, there may be mistakes in the MATLAB simulation and Arduino code that may be contributing to the poor results of implementation of the Kalman filter and full order observer. These mistakes in combination with model inconsistencies with the real-system need to be corrected in future iterations of this project.

The MATLAB and Arduino Project files can be found here.