In this paper, a new machine learning framework is developed for complex system control, called parallel reinforcement learning. To overcome data deficiency of current data-driven algorithms, a parallel system is buil...In this paper, a new machine learning framework is developed for complex system control, called parallel reinforcement learning. To overcome data deficiency of current data-driven algorithms, a parallel system is built to improve complex learning system by self-guidance. Based on the Markov chain(MC) theory, we combine the transfer learning, predictive learning, deep learning and reinforcement learning to tackle the data and action processes and to express the knowledge. Parallel reinforcement learning framework is formulated and several case studies for real-world problems are finally introduced.展开更多
In this paper the Hausdorff measure of sets of integral and fractional dimensions was introduced and applied to control systems. A new concept, namely, pseudo-self-similar set was also introduced. The existence and un...In this paper the Hausdorff measure of sets of integral and fractional dimensions was introduced and applied to control systems. A new concept, namely, pseudo-self-similar set was also introduced. The existence and uniqueness of such sets were then proved, and the formula for calculating the dimension of self-similar sets was extended to the pseudo-self-similar case. Using the previous theorem, it was shown that the reachable set of a control system may have fractional dimensions. It is expected that as a new approach the geometry of fractal sets will be a proper tool to analyze the controllability and observability of nonlinear systems.展开更多
基金supported in part by the National Natural Science Foundation of China(61503380)the Natural Science Foundation of Guangdong Province,China(2015A030310187)
文摘In this paper, a new machine learning framework is developed for complex system control, called parallel reinforcement learning. To overcome data deficiency of current data-driven algorithms, a parallel system is built to improve complex learning system by self-guidance. Based on the Markov chain(MC) theory, we combine the transfer learning, predictive learning, deep learning and reinforcement learning to tackle the data and action processes and to express the knowledge. Parallel reinforcement learning framework is formulated and several case studies for real-world problems are finally introduced.
基金This work was supported in part by Laboratory of MADIS
文摘In this paper the Hausdorff measure of sets of integral and fractional dimensions was introduced and applied to control systems. A new concept, namely, pseudo-self-similar set was also introduced. The existence and uniqueness of such sets were then proved, and the formula for calculating the dimension of self-similar sets was extended to the pseudo-self-similar case. Using the previous theorem, it was shown that the reachable set of a control system may have fractional dimensions. It is expected that as a new approach the geometry of fractal sets will be a proper tool to analyze the controllability and observability of nonlinear systems.
