Abstract
This paper is one of the researches to countermeasure against disaster. This paper treats the short-term future prediction of gas concentration in a building for safer evacuation. Moreover, in order to treat disaster, such as fire, failure of sensor installed in a building is assumed. Therefore, a fault-tolerant system against sensor's failure is designed. First, a concept of propagation coefficient is introduced, and a gas concentration model is constructed by propagation coefficient and amount of gas generation. And, in order to estimate these values, we apply Kalman Filter to this model. As this paper should consider sensor's failure, the fault-tolerant Kalman Filter is proposed. Then, by using the estimated result, the short-term future of gas concentration is predicted. Finally, we show experimental results to analyze effectiveness of the proposed method.
