A Novel Smooth Variable Structure Smoother for Robust Estimation

Adaptive Lag Smoother for State Estimation

Fixed-lag smoothing has been used across different disciplines for offline analysis in many applications. With rising computational power and parallel processing architectures, fixed-lag smoothers are increasingly integrated into online processing system with small delays. This delay is directly related to the lag-length used in system design, which needs to be chosen appropriately. In this work, an adaptive approach is devised to choose an appropriate lag-length that provides a good trade-off between accuracy and computational requirements. The analysis shown in this paper for the error dynamics of the fixed-lag smoother over the lags helps in understanding its saturation over increasing lags. In order to provide the empirical results, simulations are carried out over a second-order Newtonian system, single-axis attitude estimation, Van der Pol's oscillator, and three-axis attitude estimation. The simulation results demonstrate the performance achieved with an adaptive-lag smoother as compared to a fixed-lag smoother with very high lag-length.

An Improved Smooth Variable Structure Filter for Robust Target Tracking

As a new-style filter, the smooth variable structure filter (SVSF) has attracted significant interest. Based on the predictor-corrector method and sliding mode concept, the SVSF is more robust in the face of modeling errors and uncertainties compared to the Kalman filter. Since the estimation performance is usually insufficient in real cases where the measurement vector is of fewer dimensions than the state vector, an improved SVSF (ISVSF) is proposed by combining the existing SVSF with Bayesian theory. The ISVSF contains two steps: firstly, a preliminary estimation is performed by SVSF. Secondly, Bayesian formulas are adopted to improve the estimation for higher accuracy. The ISVSF shows high robustness in dealing with modeling uncertainties and noise. It is noticeable that ISVSF could deliver satisfying performance even if the state of the system is undergoing a sudden change. According to the simulation results of target tracking, the proposed ISVSF performance can be better than that obtained with existing filters.

Ca50 estimation based on Neural Network and smooth variable structure filter

CA50 and IMEP (indicated mean effective pressure) are two critical parameters indicating the combustion process and work output. CA50 and IMEP are playing important roles in advanced combustion modeling and control for compression-ignition and spark-ignition engines. However, both CA50 and IMEP are calculated by using the in-cylinder pressure trace, the failure or aging/wearing of cylinder pressure sensors can significantly damage the accuracy of CA50 and IMEP. In this paper, the effects of different measurement error sources, especially the pressure offset and thermal shock, on CA50 and IMEP are analyzed. The results show that thermal shock reduces the IMEP calculation accuracy, while the CA50 is sensitive to all measurement errors. To improve the accuracy and reliability of CA50 calculation, a neural network was trained to capture the CA50 by utilizing other in-cylinder information, including the temperature, pressure, oxygen concentration, injection mass as well as injection time. By treating the CA50 as "dummy" state, smooth variable structure filter is used to integrate the CA50 calculated from in-cylinder pressure and the neural network. The experimental results validate the accuracy and robustness of the proposed method.