Development of Analysis and Design Methods for Dynamical Systems Based on Algorithm Models

本研究的目的是通过将动态系统的仿真器作为其模型来开发动态系统的分析和设计方法，即基于算法描述的模型来开发分析和设计方法。为了使问题具体化，本文选取了我们一直追求的几个非线性系统和混杂系统作为研究对象。对于这些系统的一些分析和设计问题，我们开发了基于它们的仿真器，即它们的算法模型来推导解决方案的方法。作为另一种方法，我们还开发了一些方法，通过引入强化学习方法来获得一些设计问题的解，这些方法是通过学习获得模型的方法。

The purpose of this research is to develop analysis and design methods for dynamical systems by regarding their simulators as their models, that is to say, to develop analysis and design methods based on the models described by algorithms. In order to make the purpose specific issues, several nonlinear systems and hybrid systems which we have been pursuing so for are chosen as the targets of the research. For some analysis and design problems of those systems we develop methodologies to derive solutions based on their simulators, that is, their algorithm models. As another approach we also developed methodologies to derive solutions for some design problems by introducing reinforcement learning methods, which are those by acquiring their models through learning.