Abstract
Abstract
The use of self-organizing fuzzy logic controllers (SOFLCs) in high-speed multi-variable systems has been largely limited by the high number of rules generated, and by the application specific nature of the learning process. This paper concerns the development of a more generically applicable form of an SOFLC that uses a limited rule base of predetermined size, resulting in improved generalization properties and a reduction in the processing time. A simulation study on a four-valve water hydraulic actuator for a subsea robotic arm shows how this method can be applied to system modelling, the resulting 48-rule fuzzy model then providing the necessary process model for training the SOFLC. Using this fuzzy model the SOFLC was able to tune a set of 12 rules to control the actuator and then to adapt these rules to compensate for a simulated leak of hydraulic fluid.
Get full access to this article
View all access options for this article.