| 논문명 |
비선형필터를 이용한 역모델링 및 부하 추정 / Inverse modeling and load estimation using nonlinear filters |
| 주제어 |
확장 칼만 필터 ; 파티클 필터 ; 역모델링 ; 부하 추정 ; extended Kalman filter ; particle filter ; inverse modeling ; load estimation |
| 요약2 |
The Kalman filtering can effectively estimate state variables of the noisy system using a mathematical model, statistical property of measurement noise, and uncertainty in the mathematical model. However, the true value of the physical properties cannot be known due to epistemic uncertainty and technical difficulties. It signifies that the model parameters are uncertain, and thus the estimation performance is worsen and unreliable. It is one of the cons of the Kalman filter. This can be handled with non-linear filter(s) which estimates state variables and uncertain model parameters simultaneously. With this in mind, this paper explores estimation performance of the two nonlinear filters, extended Kalman filter and particle filter. A simple room model was developed for a virtual experiment and a series of experiments was conducted to verify the estimation performance under uncertain model parameters and trajectory of a system input (heat generation). The preliminary result shows that the particle filter outperforms the extended Kalman filter in terms of robustness and accuracy as well as computational cost. |
