The following information was submitted:
Transactions: WSEAS TRANSACTIONS ON CIRCUITS AND SYSTEMS
Transactions ID Number: 42-594
Full Name: Hongmei Li
Position: Ph.D. Candidate
Age: ON
Sex: Female
Address: Dept. of Electrical Engineering, Room 1-347 Shanghai Jiao Tong University 800 Dongchun Road, Minhang District Shanghai 200240, P. R. China
Country: CHINA
Tel:
Tel prefix:
Fax:
E-mail address: mei740126@yahoo.com.cn
Other E-mails:
Title of the Paper: reinforcement learning with prior knowledge approach used in AGC regulation
Authors as they appear in the Paper:
Email addresses of all the authors:
Number of paper pages: 10
Abstract: Conventional automatic generation control (AGC) systems had been developed, based on the classical linear control theory . So under the case of condition, it is impossible to considering the inherent non-linear characteristics of generator unit for an engineer, it is very difficult to determine a correct and accuracy value of the integrator gain due to the impact from the non-linear of generator unit and power grid, so the regulation results of generator unit is frequently overshoot or insufficiently. In this paper, PI controller is replaced by reinforcement-learning controller, non-linearity AGC was discretized as a Markov decision process. Area Control Error (ACE) used as state variant of the system is obtained by Q-learning. Furthermore, to improve learning efficiency, the information of AGC environment was translated into prior knowledge of Q-learning. In order to verify that the method applied in this paper is available, simulation result obtained by this AGC !
model is tested for indicate that it complies with CPS standard or not. Comparing to the standard value with the simulation results, we may conclude that the result is satisfied.
Keywords: Automatic Generation Control (AGC); integral gain; Q-learning; prior knowledge; Fuzzy Integrated Decision-making (FID)
EXTENSION of the file: .doc
Special (Invited) Session:
Organizer of the Session:
How Did you learn about congress:
IP ADDRESS: 114.93.42.192
