English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Nonlinear model predictive control (NMPC) requires accurate andcomputationally efficient plant models. Our previous work has shown that theclassical compartmentalization model reduction approach for distillationcolumns can be enhanced by replacing parts of the system of equations byartificial neural networks (ANNs) trained on offline solved solutions toimprove computational performance. In real-life applications, the absence of ahigh-fidelity model for data generation can, however, prevent the deployment ofthis approach. Therefore, we propose a method that utilizes solely plantmeasurement data, starting from a small initial data set and then continuouslyadapting to newly measured data. The efficacy of the proposed approach isexamined in silico for a distillation column from literature. To this end, wefirst adjust our reduced hybrid mechanistic/data-driven modeling approach thatoriginally builds on compartmentalization to a stage-aggregation procedure,tailoring it for the application within the adaptive learning framework.Second, we apply an adaptive learning algorithm that trains the ANNs replacingthe stationary stage-to-stage calculations on newly available data. We applythe adaptive learning of the hybrid model within a regulatory NMPC frameworkand conduct closed-loop simulations. We demonstrate that by using the proposedmethod, the control performance can be steadily improved over time compared toa non-adaptive approach while being real-time applicable. Moreover, we showthat the performance when using either a model trained on excessive amounts ofoffline generated data or the original high-fidelity model can be approached inthe limit.

Adaptive Learning of Hybrid Models for Nonlinear Model Predictive Control of Distillation Columns