No abstract

This work presents an adaptive weighting input estimation algoriJhm that efficiently and robustly on-line estimates time-varied thermal unknowns. While providing for the adaptivUy, the KtJlman filter allows us to derive a regression equetion between the bills innovation and the thermal unknown. Based on this regression model, a recursive least-squares estimator weighting by an adaptive forgetting factor is proposed to extract the unknowns that are defined as the inputs. The maximum-likelihood-type estimator (M estimator) combining the Huber psi function is used to construct the adaptive weighting forgetting factor as a function of biased innovation at each time step, thereby allowing us to estimate the unknown in a system involving measurement noise, modeling error, and unpredictable time-varying cJuznges of the unknowns. In addition, the superior capabilities of the proposed algorithm are demonstrated in several time-uarying estimate cases and two benchmark performance tests in one-dimensional inverse heat conduction problems. Also presented herein are quantitative performance test comparisons of the proposed algorithm with five inverse methods. Finally, the proposed algoriJhm simply upgrodes the conventional input estimation approach, making it appropriate for implementation purposes.

No abstract