Nonlinearity performance analysis of Gaussian weighted adaptive residual unscented particle filter

Authors

  • Dr P Sudhakar , Dr Ramana Rao, P Gopi Krishna, N J Rama Krishna

Abstract

A nonlinear problem in any area is usually much more difficult to deal with than a linear one. And
the difficulty increases with the level of nonlinearity (LON). Many techniques have been developed that
attempt to deal with this issue, including the development of various types of filters. Resolving the viability
of these methods in nonlinear circumstances is not easy. Testing should be performed with a known level
of nonlinearity. This is essential for reliable evaluation of the viability of nonlinear lessening techniques. In
this work, NEES technique was introduced because it provides useful nonlinear measurements for different
scenarios. At first, adaptive filters are applied for three different nonlinear paths (high, medium and low)
after that these filters are tested using NEES technique and RMSE parameter to measure how nonlinear the
filters are. Results show that GWARUPF gives better performance even for high degree of nonlinearities
comparatively. To elevate the performance of nonlinear filters in terms of divergence, RMSE and NEES
are applied which measures the nonlinearity of filters. When there is a deviation in RMS position error,
representing that there is nonlinearity, the values of NEES will be very high.

Published

2020-05-30

Issue

Section

Articles