The five-game clash pitting man against machine is over,with Google’s artificial intelligence program winning the series(一组比赛).The program-called AlphaGo-took four of five games against Korean Lee Sedol(李世石),a...The five-game clash pitting man against machine is over,with Google’s artificial intelligence program winning the series(一组比赛).The program-called AlphaGo-took four of five games against Korean Lee Sedol(李世石),an18-time world champion of the board game Go(围棋).展开更多
Aimed at the lack of self-tuning PID parameters in conventional PID controllers, the structure and learning algorithm of an adaptive PID controller based on reinforcement learning were proposed. Actor-Critic learning ...Aimed at the lack of self-tuning PID parameters in conventional PID controllers, the structure and learning algorithm of an adaptive PID controller based on reinforcement learning were proposed. Actor-Critic learning was used to tune PID parameters in an adaptive way by taking advantage of the model-free and on-line learning properties of reinforcement learning effectively. In order to reduce the demand of storage space and to improve the learning efficiency, a single RBF neural network was used to approximate the policy function of Actor and the value function of Critic simultaneously. The inputs of RBF network are the system error, as well as the first and the second-order differences of error. The Actor can realize the mapping from the system state to PID parameters, while the Critic evaluates the outputs of the Actor and produces TD error. Based on TD error performance index and gradient descent method, the updating rules of RBF kernel function and network weights were given. Simulation results show that the proposed controller is efficient for complex nonlinear systems and it is perfectly adaptable and strongly robust, which is better than that of a conventional PID controller.展开更多
The expert system MUST (Mining Under Structures) shown in this paper and established by the authors is a preliminary expert system to solve the policy-making problems for mining under structures by means of computers ...The expert system MUST (Mining Under Structures) shown in this paper and established by the authors is a preliminary expert system to solve the policy-making problems for mining under structures by means of computers instead of humanbeing. Based on the experience of relative experts,the authors established a knowledge base about the minings under structures,researched into reasonable method to simulate thinking processes of human experts when they are solving the problems, established the network of an expert system and named it ' MUST system' . MUST system uses the method of the structural system analysis approach. A kind of methods of Turbo Prolog and Fortran 77 language alternations is designed to meet the needs of exchange information within the MUST system. Based on this kind of methods MUST system has been constructed and realised on IBM-PC computer. For verifying the correctness, suitability and reliablity of MUST system,some practical examples of minings under structures were tentatively solved using MUST system,whose results are satisfactory.展开更多
文摘The five-game clash pitting man against machine is over,with Google’s artificial intelligence program winning the series(一组比赛).The program-called AlphaGo-took four of five games against Korean Lee Sedol(李世石),an18-time world champion of the board game Go(围棋).
基金Projects 0601033B supported by the Science Foundation for Post-doctoral Scientists of Jiangsu Province, 0C4466 and 0C060093the Scientific and Technological Foundation for Youth of China University of Mining & Technology
文摘Aimed at the lack of self-tuning PID parameters in conventional PID controllers, the structure and learning algorithm of an adaptive PID controller based on reinforcement learning were proposed. Actor-Critic learning was used to tune PID parameters in an adaptive way by taking advantage of the model-free and on-line learning properties of reinforcement learning effectively. In order to reduce the demand of storage space and to improve the learning efficiency, a single RBF neural network was used to approximate the policy function of Actor and the value function of Critic simultaneously. The inputs of RBF network are the system error, as well as the first and the second-order differences of error. The Actor can realize the mapping from the system state to PID parameters, while the Critic evaluates the outputs of the Actor and produces TD error. Based on TD error performance index and gradient descent method, the updating rules of RBF kernel function and network weights were given. Simulation results show that the proposed controller is efficient for complex nonlinear systems and it is perfectly adaptable and strongly robust, which is better than that of a conventional PID controller.
文摘The expert system MUST (Mining Under Structures) shown in this paper and established by the authors is a preliminary expert system to solve the policy-making problems for mining under structures by means of computers instead of humanbeing. Based on the experience of relative experts,the authors established a knowledge base about the minings under structures,researched into reasonable method to simulate thinking processes of human experts when they are solving the problems, established the network of an expert system and named it ' MUST system' . MUST system uses the method of the structural system analysis approach. A kind of methods of Turbo Prolog and Fortran 77 language alternations is designed to meet the needs of exchange information within the MUST system. Based on this kind of methods MUST system has been constructed and realised on IBM-PC computer. For verifying the correctness, suitability and reliablity of MUST system,some practical examples of minings under structures were tentatively solved using MUST system,whose results are satisfactory.
