System identification using Extended Kalman Filter

System identification is getting more intensive from researcher to develop an algorithm with work efficiently and more accurate. Many algorithm have been proposed to do an estimation process such as Lavemberg-Marquardt (LM), Orthogonal Least Square (OLS), Recursive Prediction Error (RPE) and Modified Recursive Prediction Error (MRPE). In this project, the model look deeper on Extended Kalman Filter (EKF) based on their advantages and specialty compared to the other technique. Basically, Extended Kalman Filter (EKF) generally was known as the optimal estimator for a dynamic system. Then, this project to do an offline estimation of several data that has been selected to execute, offline estimation means that the estimation process operate with data are provided. Besides, Extended Kalman Filter (EKF) algorithm was selected in this project as an algorithm for offline estimation data purposes. In order to evaluate the performance of the EKF learning algorithm, the proposed algorithm validation were analyzed using model validation methods as a checker such as One Step Ahead (OSA) and correlation coefficient (R2). The EKF algorithm performance was compared with Recursive Least Square (RLS) estimation algorithm as a comparison algorithm performance. All the coding simulation and the results analyzed was done using Matlab programming software.