In this paper, we introduce a novel reinforcement learning(RL) scheme for linear continuous-time dynamical systems. Different from traditional batch learning algorithms,an incremental learning approach is developed, w...In this paper, we introduce a novel reinforcement learning(RL) scheme for linear continuous-time dynamical systems. Different from traditional batch learning algorithms,an incremental learning approach is developed, which provides a more efficient way to tackle the on-line learning problem in realworld applications. We provide concrete convergence and robust analysis on this incremental-learning algorithm. An extension to solving robust optimal control problems is also given. Two simulation examples are also given to illustrate the effectiveness of our theoretical result.展开更多
A suit of online self-adapting control （OSAC） approach has been developed to predict and optimize annealing craft system. The approach consists of three critical parts including prediction module, self-adapting opti...A suit of online self-adapting control （OSAC） approach has been developed to predict and optimize annealing craft system. The approach consists of three critical parts including prediction module, self-adapting optimization module, and self-learning amendment module. Firstly, the prediction module and self- adapting optimization module are based on the modeling methods. The self-adapting optimization module consists of two parts including ＂reappearance of annealed process＂ and ＂optimization of subsequent annealing process＂. Secondly, the self-learning amendment module, based on furnace atmosphere, equipment performance, and compensation coefficients, is designed to improve the accuracy of optimization results. The results obtained from the proposed approach, usually finished in about 3 min, are in good agreement with the test values, such as the deviation of temperature for hot-spot and cold-spot are within 10 K, the relative errors are within 1.1%, and the accuracy of annealing for heating period is increased by using self-learning amendment module.展开更多
基金supported by the Natural Sciences and Engineering Research Council of Canada(N00892)in part by National Natural Science Foundation of China(51405436,51375452,61573174)
基金supported partially by the National Science Foundation(ECCS-1230040 and ECCS-1501044)
文摘In this paper, we introduce a novel reinforcement learning(RL) scheme for linear continuous-time dynamical systems. Different from traditional batch learning algorithms,an incremental learning approach is developed, which provides a more efficient way to tackle the on-line learning problem in realworld applications. We provide concrete convergence and robust analysis on this incremental-learning algorithm. An extension to solving robust optimal control problems is also given. Two simulation examples are also given to illustrate the effectiveness of our theoretical result.
基金Supported by the Specialized Research Project of WuhanIron and Steel Corporation (20050038)
文摘A suit of online self-adapting control （OSAC） approach has been developed to predict and optimize annealing craft system. The approach consists of three critical parts including prediction module, self-adapting optimization module, and self-learning amendment module. Firstly, the prediction module and self- adapting optimization module are based on the modeling methods. The self-adapting optimization module consists of two parts including ＂reappearance of annealed process＂ and ＂optimization of subsequent annealing process＂. Secondly, the self-learning amendment module, based on furnace atmosphere, equipment performance, and compensation coefficients, is designed to improve the accuracy of optimization results. The results obtained from the proposed approach, usually finished in about 3 min, are in good agreement with the test values, such as the deviation of temperature for hot-spot and cold-spot are within 10 K, the relative errors are within 1.1%, and the accuracy of annealing for heating period is increased by using self-learning amendment module.
