A basic smart linear Kalman filter with online performance evaluation based on observable degree

The observable degree can be used to directly explain the system filtering performance (or filtering accuracy) of Kalman filtering (KF) to some extent. The effective observable degree can not only be obtained before filtering but also be used to measure the system filtering performance. In applications, the exact knowledge of the system parameters and models is always unavailable. A basic smart Kalman filter (SKF) with online performance evaluation is proposed based on the observable degree in this paper. Since the collection of observations is limited in initial alignment with complex situations, mobile sensor networks are introduced. To improve the filtering performance with inaccuracy system parameters, the relatively optimal smart adjusting factor is iteratively selected by an optimized observable degree with autonomous learning function. The self-assessment function is also available for real-time performance evaluation. Finally, simulation examples are demonstrated to validate the proposed smart Kalman filter. (C) 2019 Elsevier Inc. All rights reserved.