Indirect Adaptive Fuzzy Observer Design

Jong-Kun Yang, Chang‐Ho Hyun, Jae‐Hun Kim, Euntai Kim, Mignon ParkJournal of Korean Institute of Intelligent Systems 2004 · DOI

Abstract

This paper proposes an alternative observation scheme: a Takagi-Sugeno (T-S) fuzzy model-based indirect adaptive fuzzy observer. Nonlinear systems are represented by fuzzy models, leveraging fuzzy logic systems as universal approximators. To estimate unmeasurable states, T-S fuzzy modeling is applied to capture the dynamics of the observation system. The adaptive fuzzy scheme estimates the parameters of the fuzzy model, allowing the observer to handle both unknown states and unknown parameters. The adaptive laws are derived using Lyapunov theory and the Lipschitz condition. The performance of the proposed method is validated through simulation on an inverted pendulum on a cart.

Introduction

In control engineering, "states"—such as the exact velocity of a motor or the internal temperature of a chemical reactor—are often unmeasurable due to sensor costs or physical constraints. Observers are mathematical algorithms used to estimate these hidden states from available inputs and outputs. However, traditional observers typically require a precise mathematical model. When systems are highly nonlinear or contain unknown parameters, standard approaches often fail. This paper introduces a novel indirect adaptive fuzzy observer that "learns" system parameters on the fly, providing a robust framework for state estimation in complex nonlinear environments.

Methodology

The proposed architecture integrates several key concepts from control theory to build a flexible and stable observer.

T-S Fuzzy Modeling and Observer Structure

The nonlinear system is approximated using a Takagi-Sugeno (T-S) fuzzy model, which represents the system as a set of linear "IF-THEN" rules. This allows a complex nonlinear problem to be treated as a collection of simpler linear subsystems. The observer's structure is designed to mimic this fuzzy representation, as shown in the system architecture below.

Block diagram of the proposed indirect adaptive fuzzy observer system.

Adaptive Laws and Stability

An indirect adaptive law is employed to estimate the actual parameters of the fuzzy model rather than just control gains. As these parameters are updated, the observer's accuracy improves. To ensure that the estimation error converges to zero, the authors utilize Lyapunov Stability Theory to derive the update laws. Additionally, a Lipschitz condition is applied to bound the system's nonlinearities, ensuring stability even during rapid state transitions.

Simulation and Results

The algorithm was evaluated using a numerical simulation of an inverted pendulum on a cart—a classic benchmark for unstable, nonlinear control problems. The simulation demonstrated high performance in both state tracking and parameter adaptation.

Key Results

  • State Tracking: The estimated states converged to the true states rapidly, demonstrating the observer's high fidelity.
  • Error Convergence: Observation errors dropped toward zero almost immediately after the start of the simulation.
  • Parameter Adaptation: The indirect adaptive law successfully adjusted the fuzzy model parameters to match the pendulum's dynamics, proving the observer's capability to handle unknown internal constants.
Comparison between the original state and the observed state over time.

Discussion and Conclusion

The core innovation of this work is the use of fuzzy systems as a structural framework for an observer. By treating fuzzy logic as a universal approximator, the researchers created a system capable of simultaneous state and parameter estimation. While the current formulation focuses on Single-Input Single-Output (SISO) systems, it provides a strong foundation for future extensions to Multiple-Input Multiple-Output (MIMO) systems. Future work may also address sensitivity to high-frequency measurement noise and the computational scaling of fuzzy rules for real-time embedded applications.

Summary updated

Abstract

This paper proposes an alternative observation scheme, T-S fuzzy model based indirect adaptive fuzzy observer. Nonlinear systems are represented by fuzzy models since fuzzy logic systems are universal approximators. In order to estimate the unmeasurable states of a given nonlinear system, T-S fuzzy modeling method is applied to get the dynamics of an observation system. T-S fuzzy system uses the linear combination of the input state variables and the modeling applications of them to various kinds of nonlinear systems can be found. The adaptive fuzzy scheme estimates the parameters comprising the fuzzy model representing the observation system. The proposed indirect adaptive fuzzy observer based on T-S fuzzy model can cope with not only unknown states but also unknown parameters. In the process of deriving adaptive law, the Lyapunov theory and Lipchitz condition are used. To show the performance of the proposed observation method, it is applied to an inverted pendulum on a cart.
BibTeX
@article{yang2004indirect,
  author = {Jong-Kun Yang and Chang‐Ho Hyun and Jae‐Hun Kim and Euntai Kim and Mignon Park},
  title = {{Indirect Adaptive Fuzzy Observer Design}},
  journal = {Journal of Korean Institute of Intelligent Systems},
  year = {2004},
  doi = {10.5391/jkiis.2004.14.7.927},
}
Figure