VI11 CONCLUSION
This paper presents a variable structure vector control
scheme whereby the control structure can vary from either a
purely fuzzy control to a purely neural-network control
depending on the nature of the input commands. Thus it
combines the benefits of the neural network - ability‘to
learn and predict the next command, and the benefits of the
fuzzy control - the ability to act more quickly to
unpredictable disturbances. The simpler competitive
learning technique is used for the neural network in order to
simplify the network structure and to reduce the leaming
time. The experimental results shows that this type of
control can produce good command tracking capability.
However the fuzzy control structure and the neural network
structure used are not optimized. Performance of the drive
can be improved if the optimized parameters these structures
can be determined.