Testing an Algorithm for Processing Delayed and Non-delayed Measurements

摘要

This paper addresses this problem of state estimation when measurements are provided by two different types of sensors. One of the sensors is characterized by a relatively fast sampling rate with a small time delay, and the other sensor is assumed to be more accurate from a measurement noise perspective but has a slower sampling rate and a larger time delay. This paper contains a formulation of the estimation problem in a Kalman filtering setup, and presents two different solutions. The first solution is optimal and uses a combination of smoothing and filtering. In the second solution the smoothing stage was eliminated, making it easier to implement at the expense of optimality. Simulation results for an example problem containing a two degrees-of-freedom plant are also presented. These results are compared with a solution that uses an “intuitive”, but incorrect, approach to blending the sensors measurements. In addition, the sensitivity of the solutions to sensor model in the proposed Kalman filter setup is also considered. The results show that there is indeed a significant benefit to the proposed optimal approach for blending of the sensor measurements. In addition, the analysis showed that the setup exhibits good disturbance rejection of the sensor model mismatch.