Probabilistic Approach to Analysis and Synthesis of Resilient Control Systems

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K18088
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Control engineering/System engineering
• Research Institution: Osaka University
• Principal Investigator: WADA Takayuki 大阪大学, 情報科学研究科, 准教授 (60599207)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 制御工学 / システム工学 / 数理工学 / アルゴリズム

Outline of Final Research Achievements

Probabilistic approaches to realization of resilient control systems were considered. We first provided a stopping rule of stochastic approximation procedures which is a typical method for solving unknown nonlinear equations or optimization based on its measurement of residual of equations or objective function value. A key idea is to provide a relationship between the number of iterations and estimation error of a solution candidate.
Furthermore, an identification method which is called Leave-out Sign-dominant Correlation Regions (LSCR) algorithm is considered. This method finds a non-asymptotic confidence region for the parameters of a plant model. We proposed a construction of a connected confidence region for the parameters of an AR models. A key idea is to employ the prediction error of model output directly. Since the error function is monotone with respect to each parameter, it always leads to a connected confidence region for a class of the models.

Free Research Field

制御工学