![]() |
|
|||
![]() |
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
![]() |
|
![]() |
[an error occurred while processing this directive]
17.7 Review and PreviewA fuzzy system capable of controlling a helicopter effectively and aggressively was successfully developed using a genetic algorithm to discover the fuzzy rules. Quality performance was demonstrated in both simulations and actual flight tests. The fuzzy controller architecture developed is general enough to be applicable to a variety of rotorcraft. Moving the controller to a new helicopter simply requires discovering appropriate fuzzy rules with the genetic algorithm. This chapter has addressed one of the issues that can occur in the development of real-world fuzzy systems: how does one handle large numbers of condition variables yet keep the rule set from becoming too large? In the helicopter flight control application addressed here, we have proposed exploiting domain knowledge to account for coupling effects between various variables. In this way, we have been able to limit the size of the rule base substantially. Unfortunately, this approach means that each problem is slightly different, but not so different that it cannot be used for different helicopters. However, when the rubber meets the road, isnt this always the situation? Thus far in this book we have viewed all of the problems of designing and adapting fuzzy controllers as straight optimization problems. However, there is an alternative approach that has merit and warrants mention. Thus, in the next chapter we will describe a system that employs a genetic algorithm to accomplish machine learning, a learning classifier system. Learning classifier systems evolve an entire rule set from scratch, neglecting completely rules that are unnecessary. We will describe an approach whereby learning classifier systems can be made to manipulate fuzzy production rules rather than their usual crisp rules. We will apply this alternative approach to fuzzy rule-based design to a liquid level system. ReferencesGray, F. (1953, March 17). Pulse code communication, U. S. Patent 2,632,058. Mittal, M., & Prassad, J. V. R. (1994). Modeling the UH-1H helicopter using the ARMCOP program. NASA Contract L27548DA. Saunders, G. H. (1975). Dynamics of helicopter flight. New York: John Wiley & Sons. Sugeno, M., Griffin, M. F., & Bastian, A. (1993). Fuzzy hierarchical control of an unmanned helicopter. Proceedings of the IFSA Congress, Korea Talbot, P., Tingling, B., Decker, W., & Chen, R. (1982). A mathematical model of a single main rotor helicopter for piloted simulation. NASA TM 84281.
Copyright © CRC Press LLC
![]() |
![]() |
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
![]() |
![]() |