Control of a pumped-storage hydro plant using multiple fuzzy inference systems
Item TypeConference Paper
MetadataShow full item record
The First Hydro Company’s (FHC) pumped storage hydro plant at Dinorwig in North Wales is the largest of its kind in Europe and was designed to offer a high rate of response to changes in grid frequency. However, as operational constraints have changed over the life of the plant, so have the demands placed on the operation of the machine governors. In particular, if the primary response to changes in grid frequency could be improved, this would be beneficial in terms of commercial competitiveness, particularly with regard to changes in the scheduling of generation in the U.K. It was therefore decided to investigate the possibilities of using a Fuzzy Inference System (FIS) to replace the current governor configuration at Dinorwig. A detailed model of the plant, including all mechanical, hydraulic and electrical components reproduced the major characteristics of the Dinorwig plant  and was used as the basis for the investigation. Working from the context of the Dinorwig plant, the paper surveys research into Fuzzy Control and gives a short description of previously reported work on the development of a simple Mamdani-type FIS controller for the Dinorwig plant. The simulation results show how the FIS governor offers the possibility of an improved primary response for step increases in grid loading as well as to standard injected frequency test ramps when compared with the present Proportional-Integral-Derivative (PID) governor. They also indicate how changing the size, shape and position of the FIS’s membership functions can control the response of the plant and help to reduce overshoot and settling times. Though it was noted that the single FIS could give a fairly good response over a range of loads, individual FIS’s could be ‘tuned’ to give a nearly ideal response for a given change in demand. This paper considers combining two or more FIS’s, each of which is ‘tuned’ for a specific step change in load, to improve the overall response of the enhanced governor. Simulation results are presented which indicate possible advantages of using a multi-FIS system to control the response of the hydro plant and these results are compared to those obtained from the current PID governor using normal settings as well as to the previously investigated single FIS system.
King, D.J., et al. 2002. Control of a pumped-storage hydro plant using multiple fuzzy inference systems. In: Hydro 2002: Development, Management, Performance, Kiris, Turkey, 4-7 November 2002, Conference Proceedings. pp.21-28