The brain emotional learning model can be implemented with a simple hardware and processor; however, the learning model cannot model the qualitative aspects of human knowledge. To solve this problem, a fuzzy-based emotional learning model (FELM) with structure and parameter learning is proposed. The membership functions and fuzzy rules can be learned through the derived learning scheme. Further, an emotional fuzzy sliding-mode control (EFSMC) system, which does not need the plant model, is proposed for unknown nonlinear systems. The EFSMC system is applied to an inverted pendulum and a chaotic synchronization. The simulation results with the use of EFSMC system demonstrate the feasibility of FELM learning procedure. The main contributions of this paper are (1) the FELM varies its structure dynamically with a simple computation; (2) the parameter learning imitates the role of emotions in mammalians brain; (3) by combining the advantage of nonsingular terminal sliding-mode control, the EFSMC system provides very high precision and finite-time control performance; (4) the system analysis is given in the sense of the gradient descent method.
Relation:
International Journal of Fuzzy Systems 19(3), p.942–953
