Development of decentralized data fusion algorithm with optimized kalman filter
Loading...
Date
2016-08-01
Authors
Sayed Abulhasan Quadri
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The positive virtues of data fusion technique have influenced several engineering applications to implement the technology. However, a number of challenges remain to be addressed, such as selection of appropriate algorithm, processing delay and bottleneck-memory problem. This thesis proposes a data fusion model that facilitates
selection of algorithm and recommends selected algorithm to be optimized. The model collaborates data fusion technology with algorithm engineering domain, accordingly data fusion algorithm is optimized using sophisticated technique such as functional programming to reduce the processing delay and memory usage. The model is realized in four data fusion applications such as inertial measurement unit (IMU) system, OktoKopter system, satellite data fusion and concrete structure
evaluation. In all the applications, various data fusion algorithms such as Kalman filter algorithm, factor analysis (FA) algorithm and the proposed QR-FA algorithm are compared on basis of estimation error. The proposed QR-FA algorithm is developed by introducing additional step of QR decomposition in the standard factor analysis algorithm. The algorithm with the least estimation error is selected for optimization. The results in all the applications confirm that optimization has significantly reduced execution time and memory usage of selected data fusion
algorithm.