Robust Assessment of the Degree of Influence of Identification Errors on the Dynamic Properties of a Process Model in a Multilevel Control System
Year: 2026
Type: other
Abstract: Assessing the impact of identification errors on the dynamic properties of multi-level objects under unmeasured disturbances involves analyzing deviations in model parameters and their impact on the system’s stability and accuracy. Errors at a single level (e.g., the lower control level) can cascade down the dynamics of the entire multi-level structure, especially if the disturbance cannot be measured and compensated for directly. This article analyzes the impact of identification errors on a model obtained using the least-squares method. A model of statistical errors in parameter estimation is proposed, and a probabilistically representative set of process models is defined. The estimated parameter vector is significantly amplified by small singular values, making the least-squares method extremely unreliable and impossible to obtain an accurate parameter estimate. To improve the stability of the estimate, Tikhonov’s regularization method was used, which adds a robust functional constraint to the classical least-squares correction criterion and introduces a regularization factor to regulate the balance. The results of this study, i.e., an assessment of the dynamic characteristics of the process model using a representative set, allow us to determine the upper limit of the amplitude-frequency response and obtain an estimate of the lower limit.
DOI: 10.25673/125488
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7214233856