Robustness Analysis of the Estimators for the Nonlinear System Identification.

Area of Science:

• Control Engineering
• System Identification
• Nonlinear Systems

Background:

• System identification aims to maximize dynamic information while minimizing experimental cost.
• Designing experiments that minimize deviation from normal operating conditions is crucial.
• Adaptive filtering (AF) offers a promising approach for plant model parameter estimation.

Purpose of the Study:

• To propose and evaluate an adaptive filtering (AF) scheme for plant model parameter estimation in nonlinear systems.
• To compare the performance of Minimum Error Entropy Renyi (MEER), Least Entropy Like (LEL), Least Squares (LS), and Least Absolute Deviation (LAD) estimators.
• To assess the robustness and effectiveness of different estimation methods using various excitation signals.

Main Methods:

• Adaptive filtering (AF) was employed, minimizing the error between system and plant model outputs under white noise.
• Parameter estimation was performed using the sequential quadratic programming (SQP) algorithm.
• Robustness was tested using ellipsoidal confidence regions, and effectiveness was analyzed by comparing iterations and function evaluations.

Main Results:

• The study evaluated MEER, LEL, LS, and LAD estimators for a nonlinear interacting water tanks system.
• The effectiveness of each estimator was analyzed based on computational efficiency (iterations, function evaluations).
• Numerical examples and simulation results demonstrated the performance of the proposed methods.

Conclusions:

• The paper provides a comprehensive evaluation of various estimation methods for nonlinear system identification.
• Adaptive filtering proves effective in estimating parameters for complex systems like the interacting water tanks.
• The findings contribute to optimizing experimental design and parameter estimation in control engineering.