Neuro-fuzzy tuning of Kalman filter
Petia Koprinkova‐Hristova, Kiril M. Alexiev · 2016
In the present work we upgrade the designed neuro-fuzzy approach for on-line optimal tuning of Kalman filter of a gyroscope within a Micro ElectroMechanical System (MEMS) device. In addition to the covariance matrix of measurement noise the covariance matrix of the estimated process was tuned too. Our approach consists of Adaptive Critic Design (ACD) scheme in which two actors designed as Fuzzy Rule Bases (FRBs) were tuned to adapt both covariance matrices using only information about the direction to which the estimation error changes (increase or decrease). A novel fast training dynamic neural network structure - Echo state network (ESN) - was used in the role of the critic element. Application to data collected from real MEMS demonstrated the ability of the proposed approach to tune Kalman filter and improve the quality of its estimates in changing working conditions of the MEMS in real time.