Intermittent control of unstable multivariate systems with uncertain system parameters

Area of Science:

• Control Engineering
• Systems Theory
• Applied Mathematics

Background:

• Multivariable intermittent control (MIC) integrates continuous-time optimal control with discrete state sampling.
• MIC aims to provide stability and flexibility for controlling unstable systems.
• System stability is typically independent of sample interval when using accurate model parameters.

Purpose of the Study:

• To investigate the sensitivity of MIC to inaccurate physical system parameter values.
• To determine if this sensitivity is a limitation or a beneficial characteristic for adaptive control.
• To explore the role of prediction error in identifying stable parameter combinations.

Main Methods:

• Analysis of MIC's closed-loop stability concerning model parameter inaccuracies.
• Examination of the relationship between prediction error and physical parameter values.
• Evaluation of prediction error as a trigger for state sampling events.

Main Results:

• High-dimensional parameter spaces with unstable open-loop systems present a "needle in a haystack" challenge for accurate parameter identification.
• Prediction error, sensitive to model inaccuracies, effectively identifies parameter combinations yielding stable closed-loop control and low state error.
• The sensitivity of prediction error to model inaccuracy is demonstrated as a valuable asset for adaptation.

Conclusions:

• The sensitivity of MIC to parameter inaccuracies is not solely a drawback but a potential asset for adaptive control.
• Prediction error serves as a crucial indicator for identifying optimal parameter sets and facilitating system adaptation.
• MIC's inherent sensitivity supports its rationale for combining robust control with operational flexibility.