Authors:
Carolina Maria Cardona Baron
and
Jie Zhang
Affiliation:
Newcastle University, United Kingdom
Keyword(s):
Fed-Batch Processes, Fermentation, Neural Networks, Extreme Learning Machine, Re-optimisation.
Related
Ontology
Subjects/Areas/Topics:
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Machine Learning in Control Applications
;
Neural Networks Based Control Systems
Abstract:
This paper presents using bootstrap aggregated extreme learning machine for the on-line re-optimisation
control of a fed-batch fermentation process. In order to overcome the difficulty in developing mechanistic
model, data driven models are developed using extreme learning machine (ELM). ELM has the advantage
of fast training in that the hidden layer weights are randomly assigned. A single ELM model can lack of
robustness due the randomly assigned hidden layer weights. To overcome this problem, multiple ELM
models are developed from bootstrap re-sampling replications of the original training data and are then
combined. In addition to enhanced model accuracy, bootstrap aggregated ELM can also give model
prediction confidence bounds. A reliable optimal control policy is achieved by means of the inclusion of
model prediction confidence bounds within the optimisation objective function to penalise wide model
prediction confidence bounds which are associated with uncertain predic
tions as a consequence of plant
model-mismatch. Finally, in order to deal with process disturbances, an on-line re-optimisation strategy is
developed and successfully implemented.
(More)