Online Identification of Nonlinear Stochastic Spatiotemporal System With Multiplicative Noise by Robust Optimal...

Area of Science:

• Dynamical Systems and Control Theory
• Machine Learning
• Nonlinear System Identification

Background:

• Stochastic nonlinear spatiotemporal (SST) systems with multiplicative noise are challenging to model online.
• Existing methods lack robustness and adaptive capabilities for complex dynamics.
• Accurate online identification is crucial for characterizing physical system behaviors.

Purpose of the Study:

• To propose a novel kernel method for the online identification of SST systems.
• To develop an algorithm based on robust control and model predictive control.
• To enhance the adaptive, robust, and convergent performance of online modeling.

Main Methods:

• Transformation of SST systems into partially linear kernel models (PLKMs).
• Application of reproducing kernel Hilbert space techniques.
• Development of an algorithm using model predictive control theory for online learning.

Main Results:

• The proposed method achieves adaptive, robust, and fast convergent online modeling.
• Effectively handles spatiotemporal dynamics with multiplicative noise.
• Demonstrates superior performance compared to existing learning methods in numerical studies.

Conclusions:

• The novel kernel method offers significant improvements for online identification of SST systems.
• Robust control techniques provide new insights into kernel machine learning design.
• The method facilitates better characterization of physical system properties.