2012 年 48 巻 10 号 p. 649-656
This paper deals with a fault-tolerant sensor network configuration by introducing a fault evaluation matrix and a compensation method of intermittent observation. A networked sensor system is designed by embedded local Distributed Kalman filters in each sensor, and the sensor agent has to estimate plant's state under the condition of sensor failure and intermittent observation. We propose two KF estimation algorithms which are based on a fault detection switching relies on a fault evaluation matrix and an imputation method by using estimate observation, respectively. Finally we show experimental results to analyze effectiveness of the proposed method.