Abstract A new neurofuzzy controller design algorithm using a neurofuzzy identifier is proposed. The neurofuzzy identifier identifies a fuzzy system and is used for representing the relational degrees between the reference input and output fuzzy sets. From the calculation of possibility between the neurofuzzy identifier output and the fuzzified value of the desired output, a control input for the desired output can be extracted and used for construction of a feedforward controller. In order to increse the control performace, a feedback control architecture with learning capability is also proposed. A multilayer neural network controller with the initial weight values that are obtained from learning the proportional and derivative control action is trained by an error back-propagation learning algorithm with the neural identifier for minimizing the feedback control error. Computer simulation results show that the proposed method is effective for fuzzy control of a system in the case that the initial fuzzy control rules are unknown, and the proposed feedback neurocontrol architecture is shown not only to enhance the control performaces but also to satisfy the stability criterion.
[1]
Yoshiki Uchikawa,et al.
On fuzzy modeling using fuzzy neural networks with the back-propagation algorithm
,
1992,
IEEE Trans. Neural Networks.
[2]
박철훈.
Neuro-Fuzzy Information Processing
,
1992
.
[3]
David Haussler,et al.
What Size Net Gives Valid Generalization?
,
1989,
Neural Computation.
[4]
W. Pedrycz.
Numerical and applicational aspects of fuzzy relational equations
,
1983
.
[5]
T. Yamakawa.
Fuzzy logic hardware systems
,
1989,
Symposium 1989 on VLSI Circuits.
[6]
Soo-Young Lee,et al.
Neuro-Fuzzy Controller Design Using Neuro-Fuzzy Identifier
,
1994
.
[7]
Kurt Hornik,et al.
Approximation capabilities of multilayer feedforward networks
,
1991,
Neural Networks.
[8]
Minho Lee,et al.
Neuro-Fuzzy Identifiers and Controllers
,
1994,
J. Intell. Fuzzy Syst..
[9]
Fuzzy Logic in Control Systems : Fuzzy Logic
,
2022
.
[10]
Minho Lee,et al.
A new neuro-fuzzy identification model of nonlinear dynamic systems
,
1994,
Int. J. Approx. Reason..