Fusion Algorithm of Self-adaptive Cubature Kalman Smoothing of Multi-sensor
Abstract: Aiming at
drunk-driving test system of traditional single-point automobile which has
neglected the influence of flow of in-car airflow on test precision and
accuracy, this paper puts forwards in-car drunkdriving measurement and control
method that is based on fusion technology of multi-sensors by exploration.
Based on information fusion algorithm of D-S proof theory design, main hardware
system and working mode of drunk-driving measurement and control system has
been designed; scheme design of automobile drunk-driving test and control system
that is based on multi-sensor test has been completed. When system model of
cubature Kalman smoothing is instable or abnormal, smoothing divergency will occur.
In order to solve this problem, fusion algorithm of self-adaptive cubature
Kalman smoothing is put forwarded. Noise system statistical estimator has been
designed to carry out on-line real-time estimation about statistical
characteristics of noise; smoothing process shall be modified by adopting
modified function when measurement is abnormal, so that precision of smoothing
estimation and suppression ability for smoothing divergency are both improved.
Keywords: Kalman smoothing;
Drunk-driving measurement and control; Self-adaptive; Modified function; Data
fusion
Author: Jiang Jing, Ali Kanso
Journal Code: jptkomputergg160102
