A new self-tuning method for fuzzy control

A self-tuning scheme for fuzzy logic controllers (FLC) is presented. Within the normalized phase plane, the state vector is analyzed into two components (d/sub 1/, d/sub 2/): one on the direction of the pseudoswitching line and the other on the perpendicular direction. The output scaling factors of the FLC are adjusted on-line according to the magnitudes of d/sub 1/ and d/sub 2/. The adaptation mechanism eliminates the contradictory requirements in the scaling factors resulting from different performance measures and therefore, both the responsiveness and stability, of the system are improved. The proposed self-tuning technique can cope with large parameter uncertainties and high nonlinearities. Moreover, it is model independent and easy to design. Comparison of this scheme with other adaptive methods is done via intensive computer simulations.