Higher Sampling Rate Estimation by a Kalman Filter with Dual Models : Influence for Modeling Error

概要

論文の詳細を見る
With the intention of achieving a higher sampling rate estimation, we have studied a Kalman filter with dual models, under the condition that we cannot change the measured signal sampling rate, but can increase the input signal sampling rate. We evaluate the estimation errors for the proposed Kalman filter and a conventional Kalman filter using computer simulations. The proposed Kalman filter is constituted of two models that are similar to the real plant. Moreover, we examine an influence on the estimation error that is caused by the modeling error for the system. From the results the MSE (Mean Square Error) of the estimation error for the proposed system is over 6 times as small as that for the conventional estimation. Moreover, the MSE of the estimation error for the proposed estimation keeps in very low levels and almost constant even if the modeling error changes from -30% to +30%. Therefore, it is shown that the proposed Kalman filter with dual models is more precise and robust against the modeling error than the conventional estimation.
大妻女子大学の論文