Virtual Reference Feedback Tuning for Microalgae Culture Control

Microalgae are photosynthetic microorganisms capable of fix carbon dioxide (CO2) to produce oxygen
(O2) and various metabolite types. The crop of these microorganisms represents a challenge when
controlling this type of bioprocesses. The above is due to the biological, physical, and chemical variability
of the initial operating conditions and the non-linear dynamics of this bioprocess. To this date, different
control strategies focused on the regulation of physicochemical variables such as temperature, pH, dilution
rate, and light intensity, among others, have been reported. However, these control paradigms usually
depend heavily on the mathematical model used to represent some multifactorial phenomena specific to the
bioprocess. Or, on the contrary, parametric identification models that generalize such behaviors without
being able to consider all the intrinsic and extrinsic dynamics of the bioprocess. As a result, the design of
model-based controllers that, to front parameter variation and disturbances, may perform less robustly.
Model-Free Control strategies such as Virtual Reference Feedback Tuning (VRFT) represent a novel
alternative paradigm based on virtual iterative reference generation until a system steady-state error of
close to zero is achieved. The implementation of VRFT strategy involves: 1) collection of experimental
data obtained from the open-loop plant to obtain the parameter values, 2) minimization of an objective
function obtained from the closed-loop system, and 3) linearization of the phenomenological model of the
process in its stable equilibrium region. This strategy is a tool for the robust control of bioprocesses, due to
its iterative characteristic that leads to an auto-tuning (on-line/off-line) of the controller. This paper shows
the design and implementation of the control strategy by the VRFT for a closed continuous microalgae
culture in three flat PBRs with a volume of 3L. The control strategy was developed in Python 3.9 and
Matlab 2020® and implemented on an ARM Cortex-M3 microcontroller integrated with an embedded
system. The process variables controlled were temperature and light intensity. The results showed a
follow-up to the reference values, on the one hand, a steady state temperature of 25 ℃ ± 0.625 ℃, on the
other hand, a light intensity of 100 μmol·m−2·s−1 ± 5 μmol·m−2·s−1. All this suggests that this strategy is a
good alternative to applied in the control of bioprocesses due to the characteristics of these. In the future, it
is expected that this prototype can be scale-up and taken to an industrial level to provide support for
increased culture production in less time.