Extended set-membership filter for indoor gas source localization
Huayun Shi, Yi Chai, Shanbi Wei, Youtian Xia · 2016
A novel estimation method is presented in the frame of extended set-membership filter(ESMF) for indoor gas source localization on sensor array. The noise sources of sensors are assumed unknown but bounded. Gas plume model is used to calculate localization error bound by ESMF iteratively. The estimation result is a set that contains the true location, rather than a point or points as in probabilistic estimation algorithms such as extended Kalman filter(EKF), unscented Kalman filter(UKF) etc. Thus it provides 100% confidence for the estimated source location. However, inappropriate initial value and initial ellipsoid will affect numerical stability of ESMF. Hence, we obtain an initial value by Nonlinear Least Squares Method(NLSM) in preliminary location. An initial ellipsoid of relatively small volume is calculated according to the error of preliminary location and Lagrange remainder interval analysis of measurement equation. The location with high accuracy can be got in short time. Finally, simulation is taking based on wireless electronic nose network to demonstrate the feasibility and effectiveness of the presented method.