Abstract
This paper concerns with both the system identification problem based on the environmental data and the prediction problem of the identified system state by Kalman filtering method. Furthermore, it is shown that this method can be applied to the selection of the observation sensor locations. These methods are tested by the data measured in Osaka city.
A kind of lower order model is made by applying the factor analysis to the multiple linear regression model constructed by the actual data. Kalman filtering prediction algorithm with the correlation between the system noise and the observation noise is used to predict the state of this lower order model and the state of the original model. We describe the cases when the statistics of the observation noise is known or unknown. The accuracy of the prediction presented above is compared with the persistence model by numerical examples.
