SISO nonlinear system identification using a fuzzy-neural hybrid system

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• System Identification

Background:

• Nonlinear dynamic systems often possess unknown parameters, posing challenges for accurate modeling and control.
• Traditional identification methods can be complex and computationally intensive.
• Integrating fuzzy logic and neural networks offers a promising avenue for enhanced system identification.

Purpose of the Study:

• To develop and evaluate a fuzzy-neural hybrid system for identifying nonlinear dynamic systems with unknown parameters.
• To leverage the strengths of both fuzzy systems and neural networks for improved model complexity and learning speed.
• To demonstrate the effectiveness of the proposed hybrid model for Single-Input Single-Output (SISO) dynamic systems.

Main Methods:

• A context-sensitive modular approach combining a fuzzy system (function module) and a multilayer neural network (context module).
• Decomposition of the hybrid system to reduce complexity and expedite the learning process.
• Utilizing the physical interpretability of fuzzy system parameters for incorporating prior knowledge and constraints.
• Employing the gradient descent method for parameter adjustment within the hybrid system.

Main Results:

• The fuzzy-neural hybrid system effectively identifies nonlinear dynamic systems with unknown parameters.
• The decomposed structure of the hybrid system significantly reduces complexity and accelerates the learning process.
• Simulations confirm the high effectiveness of the proposed hybrid identification models for SISO dynamic systems.

Conclusions:

• Fuzzy-neural hybrid systems offer a powerful and efficient approach for nonlinear dynamic system identification.
• The proposed model's structure facilitates practical implementation using fast, parallel devices.
• The ability to incorporate a priori knowledge enhances the robustness and accuracy of the identification process.