Management of Uncertainties and Risks
03 Evaluation of an environmental risk Project Leader
Koji Kanefuji
In order to solve an environmental problem, we have to grasp the present condition of contamination correctly. For that purpose, it is necessary to obtain data with sufficient accuracy. And also it is indispensable the technique of the statistical science for determining a sampling place and time. Statistical science offers the scientific base to the community of environmental science, building the ultramodern statistical methodology turned to solution of the environmental problem in water or the atmosphere from the obtained data.


Members
Koji Kanefuji

Fig.1

(Photo: Courtesy of Prof. Fujie, Toyohashi University of Technology)
04 Study on identification and control of processes which include nonlinear parametric models Project Leader
Yoshihiko Miyasato
Certain nonlinear parametric models, such as three-layered neural networks are known to be useful tools to approximate nonlinear components included in many physical processes. In the present research, we provided a design method of control systems for processes which include nonlinear parametric models by utilizing collaboration approach of control and estimation. The nonlinear parametric models are used to approximate nonlinear elements in controlled processes, and are updated recursively in accordance with the new data of inputs and outputs. In order to assure the stability of overall systems, the approximate errors in those nonlinear parametric models, algorithmic errors contained in the update schemes of tuning parameters, and unknown system parameters are regarded as exogenous disturbances to the processes. The resulting nonlinear control strategies are derived as semi-optimal solutions for the H-infinity control problems with those virtual disturbance. The collaboration of those estimation and control schemes stabilize the overall system and attain good system performance.


Members
Yoshihiko Miyaato

 Fig.1
| 01 | 02 | 03 | 04 | Category Index