Pith. sign in

REVIEW 4 major objections 4 minor 42 references

A reinforcement learning agent, pretrained by imitating a PID controller and fine-tuned daily, is deployed for eight days on an open photobioreactor, keeping pH close to setpoint and, in simulation, reducing control effort by 54% versus PID

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-04 22:58 UTC pith:WHPHZKT6

load-bearing objection A credible demonstration-scale RL deployment for pH control in an open photobioreactor, with an overbroad 'first' claim and a single simulation run. the 4 major comments →

arxiv 2509.06853 v1 pith:WHPHZKT6 submitted 2025-09-08 eess.SY cs.AIcs.LGcs.SY

Reinforcement learning meets bioprocess control through behaviour cloning: Real-world deployment in an industrial photobioreactor

Juan D. Gil , Ehecatl Antonio Del Rio Chanona , Jos\'e L. Guzm\'an , Manuel Berenguel This is my paper
classification eess.SY cs.AIcs.LGcs.SY
keywords reinforcement learningbehaviour cloningphotobioreactorpH controlDDPGoffline RLbioprocess controlonline fine-tuning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper claims that reinforcement learning can be moved from simulation to the real world for bioprocess control: a DDPG agent, pretrained offline on data from a PID controller and then retrained briefly each evening, regulated pH in an open 80-square-metre photobioreactor for eight consecutive days. The authors argue this is the first real-world deployment of an RL controller on such a nonlinear, disturbance-prone bioprocess. In simulation, the fine-tuned agent reduced integrated absolute error by about 8% versus PID and 5% versus offline RL, while cutting control effort by roughly 54% and 7%, respectively. The point is not that RL beats PID by a fixed margin, but that behaviour cloning gives RL a safe, cheap starting point and daily fine-tuning lets it adapt as the biology and weather drift. If the result holds, it opens a practical route for RL in other processes where online exploration is too risky or expensive.

Core claim

The central claim is that a reinforcement learning policy can be trained offline from PID-generated trajectories and then adapted online to control pH in an open photobioreactor, with the real-plant deployment demonstrated for eight days in June 2025. The authors design a POMDP whose observations combine direct measurements (temperature, irradiance, dissolved oxygen, dilution and air flows, CO2 injection) with control error and its integral, giving the agent an implicit feedforward view of disturbances; a logarithmic reward keeps gradients informative near zero error and tolerant of large transients. Simulation over three days on unseen-season data shows the fine-tuned agent (RL-FT) achievin

What carries the argument

The load-bearing mechanism is the offline–online training protocol with behaviour cloning. Offline, the DDPG actor–critic learns from two days of PID closed-loop trajectories stored as (observation, action, reward, next observation) tuples; online, at the end of each day, the agent is fine-tuned for 50 epochs against a replay buffer that starts from the historical data and continuously replaces old experiences with the day's new ones. The observation space is the second load-bearing piece: by feeding irradiance, DO, Qair, Qd, temperature, and CO2 alongside the error and error integral, the agent acquires intrinsic feedforward disturbance rejection, while a clipped integral prevents actuator-

Load-bearing premise

The claimed improvements over PID are quantified in a simulation model whose fidelity to the real reactor under the tested disturbances is assumed; the real eight-day trial reports pH deviations but does not run PID on the same days for comparison.

What would settle it

Run the published raceway model with the same training data and the three controllers, and check whether RL-FT reproduces IAE 2162.0 and CCE 140.30; alternatively, on the physical reactor, alternate days between PID and RL-FT under matched weather for at least eight days—if PID's IAE or CO2 consumption is not worse, the central performance claim fails.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Offline imitation of a nominal PID gives RL a safe initialization for bioprocesses, removing the need for risky online exploration at deployment time.
  • Daily fine-tuning with a rolling replay buffer turns a fixed offline policy into an adaptive one, which is what allowed the agent to handle conditions absent from the training days.
  • In simulation, the fine-tuned agent cut cumulative control effort by 54% versus PID, implying lower CO2 consumption and operating cost.
  • The POMDP observation design—measurable disturbances plus error and error-integral—is a reusable recipe for pH control in open photobioreactors and possibly other culture systems.
  • Real-plant operation over eight days, including sensor recalibration and communication losses, suggests the method degrades gracefully under operational disruptions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The 8% IAE and 54% CCE margins come from a simulator; the real deployment has no concurrent PID baseline, so the quantitative claim is untested head-to-head on the physical reactor.
  • The daily 50-epoch fine-tuning is the likeliest active ingredient for the observed adaptation; ablating the fine-tuning schedule on the real plant would separate adaptation gains from the behaviour-cloned base policy.
  • The log reward function's role could be isolated by retraining the same agent with quadratic error reward on the same dataset; if IAE holds, the reward choice is less critical than the observation design.
  • A natural next test is transferring this offline–online recipe to a different bioprocess (e.g., tubular photobioreactor or wastewater treatment) to see whether the first-real-deployment claim generalizes beyond raceway PBRs.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper proposes a hybrid offline-online reinforcement learning (RL) controller for pH regulation in an open raceway photobioreactor. The agent is first trained offline with deep deterministic policy gradient (DDPG) on trajectories generated by a nominal PID controller, then fine-tuned online once per day using recent operational data. A POMDP observation space includes measured disturbances (irradiance, DO, dilution and air flows), time information, and control variables. Simulation results on a validated model compare PID, an RL agent without fine-tuning, and the proposed RL-FT agent, reporting about 8% IAE improvement over PID and 54% reduction in control effort. The method was deployed for eight days in June 2025 on an 80 m^2 raceway reactor, with qualitative evidence of disturbance rejection and adaptation. The paper claims this is the first real-world deployment of an RL-based controller for a complex, highly nonlinear, multi-disturbed system, and the first such application to a bioprocess.

Significance. If the claims are properly supported, the paper is a meaningful step toward practical RL-based bioprocess control. Its strengths include a realistic POMDP formulation, use of existing PID-generated data to avoid risky online exploration, explicit hyperparameter reporting, a publicly available plant model, and an actual multi-day field deployment with operational events (sensor recalibration, communication loss) disclosed. The demonstrated ability to fine-tune daily and reduce CO2 control effort is practically relevant. However, the central novelty claim is overstated as written, and the real-plant evidence is descriptive rather than quantitative, so the significance of the field deployment is currently not established at the level the conclusions assert.

major comments (4)
  1. [Section 1] The claim 'The most significant achievement of this work is demonstrating, for the first time to our knowledge, the successful real-world deployment of a RL–based approach for controlling a complex, highly nonlinear, and multi-disturbed system' is internally inconsistent with the cited literature. The same section cites Seo et al. (2025), 'Implementation of reinforcement learning for enhanced pressure control in a 190,000-barrel crude distillation unit: The first full-scale commercial deployment.' A crude distillation unit is a complex, nonlinear, multi-disturbed industrial process; if that deployment counts, the broad 'first' claim is false. The narrower claim in the abstract and Section 3—first application to a nonlinear and disturbance-prone bioprocess—may be defensible, but the introduction's strongest claim must be revised accordingly.
  2. [Section 4.3] The real-world validation is reported only descriptively. No IAE, CCE, or other quantitative metrics are computed for the 8-day deployment, and there is no concurrent PID baseline or repeated trial. Statements such as 'confirmed the robustness and reliability of the proposed approach' (abstract) and the conclusion that the algorithm's efficacy was 'validated' go beyond what the data shown can support. The paper should either provide quantitative performance indices for the real plant, compare against a PID run under comparable conditions (even if not concurrent), or explicitly frame the field test as a feasibility demonstration with its limitations.
  3. [Section 3.2 / Algorithm 2] The title and abstract claim the approach works 'through behaviour cloning,' and Section 1 defines BC as a supervised-learning method that trains a policy to imitate expert behaviors. However, Algorithm 2 does not perform BC: it trains a DDPG agent offline using the standard actor-critic policy-gradient objective on PID-generated transitions. Offline RL from expert data is not equivalent to behavior cloning. This mismatch affects the methodological identity of the contribution and should be corrected, either by adding an explicit BC pretraining stage or by reframing the text as offline RL initialized from expert demonstrations.
  4. [Table 1 / Section 4.2] The quantitative advantages of RL-FT over PID and RL rest on a single three-day test scenario. There are no error bars, no multiple random seeds, and no sensitivity analysis with respect to the test period or initial network weights. The reported differences—8% IAE and 54% CCE—could be within run-to-run variability. The simulation study should report repeated trials or otherwise justify that the comparison is statistically stable, or the claims should be softened accordingly.
minor comments (4)
  1. [Eq. (6) / Fig. 5] The text says the reward uses ε=10^6, but the reward values shown in Figs. 10 and 13 (0 to 15) and the stated maximum at zero error imply ε=10^-6. Please clarify the intended value and correct the figure/text inconsistency.
  2. [Section 4.1] The actor and critic are described as 'a recurrent neural network' but the architecture described and shown in Fig. 7 consists of fully connected layers with no recurrent connections. This is confusing and should be reworded.
  3. [Section 4.1] The expert controller is described as PID 'without the derivative term'—this is a PI controller. Also, T_i is given as 1200 [s^-1]; the units should be [s].
  4. [Abstract] The abstract's phrase 'first application of an RL-based control strategy to such a nonlinear and disturbance-prone bioprocess' would benefit from a more specific scope (e.g., open photobioreactors) or a literature check, since broader bioprocess RL deployments may exist.

Circularity Check

0 steps flagged

No significant circularity: the RL policy is trained from PID trajectories but evaluated post hoc on independent metrics, and the central claim is supported by a real-world deployment.

full rationale

The paper's derivation chain is not circular. The RL agent is trained offline on PID-generated trajectories and then fine-tuned online, but the reported IAE/CCE improvements are computed after training/fine-tuning from the resulting closed-loop behavior, not fitted as inputs to the evaluation. The reward function (Eq. 6) depends on control error only; IAE and CCE (Eq. 7) are separate post-hoc metrics, so there is no definitional equivalence between the training objective and the reported results. The simulation model is cited from prior same-group work (Nordio et al. 2024; Sánchez-Zurano et al. 2021; Rodríguez Miranda et al. 2025), but it is available as a code artifact on Zenodo and validated outdoors, and the central claim of successful real-world deployment is independently evidenced by the 8-day experiment in Section 4.3, which reports absolute pH deviations (max 0.08–0.11) without relying on the simulation model for the deployment conclusion. No uniqueness theorem, ansatz, or fitted parameter is imported from the authors' prior work to force the results. The 'first real-world deployment' statement is internally inconsistent with the cited Seo et al. (2025) full-scale commercial deployment, but that is an overclaim/correctness risk, not a circularity. Therefore the circularity score is 0.

Axiom & Free-Parameter Ledger

10 free parameters · 4 axioms · 0 invented entities

The central claim rests on the fidelity of the simulation model and on the adequacy of the observation/reward design. No new physical entities are postulated. The free parameters are RL hyperparameters and expert-controller gains, not fitted to the target performance metrics.

free parameters (10)
  • Reward epsilon = 1e6
    Chosen in Eq. (6) to set the maximum reward; affects gradient scale during offline training.
  • Discount factor = 0.9
    DDPG hyperparameter; hand-selected in Section 4.1.
  • Smoothing factor tau = 0.01
    Target-network soft update factor, Section 4.1.
  • Learning rates = 1e-4 critic, 1e-5 actor
    Adam optimizer settings, Section 4.1.
  • Offline training epochs = 4000
    Duration of offline DDPG training, Section 4.1.
  • Fine-tuning epochs per day = 50
    Daily update budget to avoid overfitting, Section 4.1.
  • Mini-batch size = 64
    Replay sampling size, Section 4.1.
  • Hidden layer width = 256
    Width of actor and critic layers, Section 4.1.
  • Sampling time Ts = 10 s
    Control interval matching the real facility, Section 4.1.
  • PID gains for expert data = Kp=-32 L/min, Ti=1200 s^-1
    Expert controller parameters from Caparroz et al. (2023), used to generate demonstrations.
axioms (4)
  • domain assumption The validated microalgae model (Nordio et al. 2024; Sanchez-Zurano et al. 2021; Rodriguez Miranda et al. 2025) accurately represents the real PBR dynamics for the tested scenarios.
    Used in Section 4.2 to generate training data and compute all quantitative comparisons; the model is external but from closely related prior work by the same group.
  • domain assumption The engineered observation space (T, I, DO, Qd, Qair, CO2, time, error, error integral) provides sufficient information about the hidden biological state to learn a stabilizing policy.
    POMDP formulation in Section 3.1.1; biomass concentration is not measured, so the integral of error is used to compensate for partial observability.
  • domain assumption The pH response follows the carbonate-equilibrium and photosynthesis mechanisms described in Section 2.1.2.
    Basis for the control problem; standard bioprocess domain knowledge.
  • domain assumption The PID controller's recorded actions constitute expert demonstrations suitable for behavior cloning and offline RL.
    Section 3.2 relies on PID-generated trajectories as the offline training dataset; this assumes the PID is a competent expert policy.

pith-pipeline@v1.3.0-alltime-deepseek · 3820 in / 3912 out tokens · 156021 ms · 2026-08-04T22:58:55.377194+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of Reinforcement learning meets bioprocess control through behaviour cloning: Real-world deployment in an industrial photobioreactor." pith.science (2026). https://pith.science/paper/WHPHZKT6

@misc{pith2026250906853,
  author       = {Pith},
  title        = {Pith review of: Reinforcement learning meets bioprocess control through behaviour cloning: Real-world deployment in an industrial photobioreactor},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WHPHZKT6}},
  note         = {Machine review of arXiv:2509.06853}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The inherent complexity of living cells as production units creates major challenges for maintaining stable and optimal bioprocess conditions, especially in open Photobioreactors (PBRs) exposed to fluctuating environments. To address this, we propose a Reinforcement Learning (RL) control approach, combined with Behavior Cloning (BC), for pH regulation in open PBR systems. This represents, to the best of our knowledge, the first application of an RL-based control strategy to such a nonlinear and disturbance-prone bioprocess. Our method begins with an offline training stage in which the RL agent learns from trajectories generated by a nominal Proportional-Integral-Derivative (PID) controller, without direct interaction with the real system. This is followed by a daily online fine-tuning phase, enabling adaptation to evolving process dynamics and stronger rejection of fast, transient disturbances. This hybrid offline-online strategy allows deployment of an adaptive control policy capable of handling the inherent nonlinearities and external perturbations in open PBRs. Simulation studies highlight the advantages of our method: the Integral of Absolute Error (IAE) was reduced by 8% compared to PID control and by 5% relative to standard off-policy RL. Moreover, control effort decreased substantially-by 54% compared to PID and 7% compared to standard RL-an important factor for minimizing operational costs. Finally, an 8-day experimental validation under varying environmental conditions confirmed the robustness and reliability of the proposed approach. Overall, this work demonstrates the potential of RL-based methods for bioprocess control and paves the way for their broader application to other nonlinear, disturbance-prone systems.

Figures

Figures reproduced from arXiv: 2509.06853 by Ehecatl Antonio del Rio Chanona, Jos\'e L. Guzm\'an, Juan D. Gil, Manuel Berenguel.

Figure 1
Figure 1. Figure 1: Real open PBR facilities, in Almería, Spain. The two reactors have identical [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: PBR physical scheme, (a) top view, (b) side view. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: 8 [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 3
Figure 3. Figure 3: External representation of the PBR system. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Reinforcement learning framework. The left diagram shows the framework for [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison of rewards function with respect to the error. The logarithmic [PITH_FULL_IMAGE:figures/full_fig_p015_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Proposed control structure. 3.2. RL algorithm training The methodology proposed in this work combines the offline training of the RL agent with subsequent online fine-tuning, and it is fully described in Algorithm 2. The offline component enables the agent to learn an initial control policy from historical data generated by an expert system, specifically a PID controller. The online fine-tuning phase, on t… view at source ↗
Figure 7
Figure 7. Figure 7: Actor and critic structure. This corresponds to Step 1 of Algorithm 2. Using these data, offline training was then conducted over 4000 epochs (Step 2 of Algorithm 2). For online deployment and fine-tuning (Step 3), communication with the real facility was established through an industrial protocol, specifically via an OPC DA server. This protocol enabled the retrieval of observations from the real PBR syst… view at source ↗
Figure 8
Figure 8. Figure 8: Flow-chart representing the implementation of the proposed methodology. In [PITH_FULL_IMAGE:figures/full_fig_p020_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Operation of the simulated PBR system using the PID controller. (a) pH in the [PITH_FULL_IMAGE:figures/full_fig_p021_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Values of the reward function during the PID control of the simulated PBR [PITH_FULL_IMAGE:figures/full_fig_p022_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Results obtained from the simulated system using the three controllers. (a) pH [PITH_FULL_IMAGE:figures/full_fig_p023_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Operation of the real PBR system using the PID controller. (a) pH in the [PITH_FULL_IMAGE:figures/full_fig_p025_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Values of the reward function during the PID control of the real PBR system. [PITH_FULL_IMAGE:figures/full_fig_p026_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Operation of the real PBR system using the proposed methodology. (a) pH in the PBR system (pH), pH reference [PITH_FULL_IMAGE:figures/full_fig_p028_14.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

42 extracted references · 36 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  3. [3]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  4. [4]

    , author Mendes, J

    author Amaro, M. , author Mendes, J. , author Matias, T. , author Ara \'u jo, R. , year 2023 . title Adaptive fuzzy generalized predictive control of pH in tubular photobioreactors on microalgae plant , in: booktitle 2023 IEEE 2nd Industrial Electronics Society Annual On-Line Conference (ONCON) , organization IEEE . pp. pages 1--6

    work page 2023

  5. [5]

    , author B gh, S

    author Blad, C. , author B gh, S. , author Kalles e, C.S. , year 2022 . title Data-driven offline reinforcement learning for HVAC -systems . journal Energy volume 261 , pages 125290

    work page 2022

  6. [6]

    , author De Bruin, T

    author Bu s oniu, L. , author De Bruin, T. , author Toli \'c , D. , author Kober, J. , author Palunko, I. , year 2018 . title Reinforcement learning for control: Performance, stability, and deep approximators . journal Annual Reviews in Control volume 46 , pages 8--28

    work page 2018

  7. [7]

    , author Guzm \'a n, J.L

    author Caparroz, M. , author Guzm \'a n, J.L. , author Berenguel, M. , author Aci \'e n, F.G. , year 2024 . title A novel data-driven model for prediction and adaptive control of pH in raceway reactor for microalgae cultivation . journal New Biotechnology volume 82 , pages 1--13

    work page 2024

  8. [8]

    , author Guzm \'a n, J.L

    author Caparroz, M. , author Guzm \'a n, J.L. , author Berenguel, M. , author Gil, J.D. , author Aci \'e n, F.G. , year 2023 . title Model reference adaptive control for pH regulation . journal Revista Iberoamericana de Autom \'a tica e Inform \'a tica Industrial (RIAI) volume 22 , pages 126--134

    work page 2023

  9. [9]

    , author Guzm \'a n, J.L

    author Caparroz, M. , author Guzm \'a n, J.L. , author Gil, J.D. , author Berenguel, M. , author Aci \'e n, F.G. , year 2025 a. title A hybrid MRAC-PI approach to regulate pH in raceway reactors for microalgae production . journal Control Engineering Practice volume 156 , pages 106191

    work page 2025

  10. [10]

    , author Soltesz, K

    author Caparroz, M. , author Soltesz, K. , author H \"a gglund, T. , author Guzm \'a n, J.L. , author Berenguel, M. , year 2025 b. title A new approach to relay-based autotuning pid controllers and their evaluation in ph control of industrial photobioreactors . journal Control Engineering Practice volume 164 , pages 106520

    work page 2025

  11. [11]

    , author Campoy-Iniesta, C

    author Castilla, M.M. , author Campoy-Iniesta, C. , author \'A lvarez, J.D. , year 2025 . title Reinforcement learning based thermal comfort control in a bioclimatic building . journal Revista Iberoamericana de Autom \'a tica e Inform \'a tica Industrial (RIAI) volume 22 , pages 146--155

    work page 2025

  12. [12]

    , author Sierla, S

    author Deng, J. , author Sierla, S. , author Sun, J. , author Vyatkin, V. , year 2023 . title Offline reinforcement learning for industrial process control: A case study from steel industry . journal Information Sciences volume 632 , pages 221--231

    work page 2023

  13. [13]

    , author Pe \ n a, J

    author Fern \'a ndez, I. , author Pe \ n a, J. , author Guzm \'a n, J. , author Berenguel, M. , author Aci \'e n, F. , year 2010 . title Modelling and control issues of ph in tubular photobioreactors . journal IFAC Proceedings Volumes volume 43 , pages 186--191

    work page 2010

  14. [14]

    , author Dewasme, L

    author Feudjio Letchindjio, C.G. , author Dewasme, L. , author Vande Wouwer, A. , year 2021 . title An experimental application of extremum seeking control to cultures of the microalgae scenedesmus obliquus in a continuous photobioreactor . journal International Journal of Adaptive Control and Signal Processing volume 35 , pages 1285--1297

    work page 2021

  15. [15]

    , author Aci \'e n, F.G

    author Guzm \'a n, J.L. , author Aci \'e n, F.G. , author Berenguel, M. , year 2021 . title Modelling and control of microalgae production in industrial photobioreactors . journal Revista Iberoamericana de Autom \'a tica e Inform \'a tica Industrial (RIAI) volume 18 , pages 1--18

    work page 2021

  16. [16]

    , author Berenguel, M

    author Guzm\'an, J.L. , author Berenguel, M. , author Rodr\'iguez-Miranda, E. , author Aci\'en, F.G. , year 2025 . title Microalgae production at industrial scale: Modelling and control challenges , in: booktitle American Control Conference 2025 (ACC25), Denver (USA), July 8-10

    work page 2025

  17. [17]

    , author H \"a gglund, T

    author Guzm \'a n, J.L. , author H \"a gglund, T. , year 2024 . title Feedforward Control: Analysis, Design, Tuning rules, and Implementation . publisher Walter de Gruyter GmbH & Co KG

    work page 2024

  18. [18]

    , author Guzm \'a n, J.L

    author H \"a gglund, T. , author Guzm \'a n, J.L. , year 2024 . title Give us pid controllers and we can control the world . journal IFAC-PapersOnLine volume 58 , pages 103--108

    work page 2024

  19. [19]

    , author Bahri, P.A

    author Isiramen, O.E. , author Bahri, P.A. , author Moheimani, N.R. , author Vadiveloo, A. , author Shayesteh, H. , author Parlevliet, D.A. , year 2022 . title Improving pH control and carbon dioxide utilisation efficiency in microalgae cultivation systems with the use of a proportional-integral+ dead-zone control strategy . journal Bioresource Technology...

    work page 2022

  20. [20]

    , author Ceballos, R.M

    author Juneja, A. , author Ceballos, R.M. , author Murthy, G.S. , year 2013 . title Effects of environmental factors and nutrient availability on the biochemical composition of algae for biofuels production: A review . journal Energies volume 6 , pages 4607--4638

    work page 2013

  21. [21]

    , author Ba, J

    author Kingma, D.P. , author Ba, J. , year 2014 . title Adam: A method for stochastic optimization . journal arXiv preprint arXiv:1412.6980

    Pith/arXiv arXiv 2014

  22. [22]

    , author Kumar, A

    author Levine, S. , author Kumar, A. , author Tucker, G. , author Fu, J. , year 2020 . title Offline reinforcement learning: Tutorial, review, and perspectives on open problems . journal arXiv preprint arXiv:2005.01643

    Pith/arXiv arXiv 2020

  23. [23]

    , author Hunt, J.J

    author Lillicrap, T.P. , author Hunt, J.J. , author Pritzel, A. , author Heess, N. , author Erez, T. , author Tassa, Y. , author Silver, D. , author Wierstra, D. , year 2015 . title Continuous control with deep reinforcement learning . journal arXiv preprint arXiv:1509.02971

    Pith/arXiv arXiv 2015

  24. [24]

    , author Fan, Y

    author Liu, K. , author Fan, Y. , author Chen, J. , year 2024 . title A model free adaptive control method based on self-adjusting PID algorithm in pH neutralization process . journal Chinese Journal of Chemical Engineering volume 76 , pages 227--236

    work page 2024

  25. [25]

    , author Kurian, V

    author Luo, Y. , author Kurian, V. , author Ogunnaike, B.A. , year 2021 . title Bioprocess systems analysis, modeling, estimation, and control . journal Current Opinion in Chemical Engineering volume 33 , pages 100705

    work page 2021

  26. [26]

    title MATLAB Optimization Toolbox Release 2023a

    author MATLAB , year 2023 . title MATLAB Optimization Toolbox Release 2023a . publisher The MathWorks, Natick, MA, USA

    work page 2023

  27. [27]

    , author Lafarga, T

    author Morillas-Espa \ n a, A. , author Lafarga, T. , author G \'o mez-Serrano, C. , author Aci \'e n, F.G. , author Gonz \'a lez-L \'o pez, C.V. , year 2020 . title Year-long production of scenedesmus almeriensis in pilot-scale raceway and thin-layer cascade photobioreactors . journal Algal Research volume 51 , pages 102069

    work page 2020

  28. [28]

    , author Rodr \' guez-Miranda, E

    author Nordio, R. , author Rodr \' guez-Miranda, E. , author Casagli, F. , author S \'a nchez-Zurano, A. , author Guzm \'a n, J.L. , author Aci \'e n, F.G. , year 2024 . title ABACO-2: A comprehensive model for microalgae-bacteria consortia validated outdoor at pilot-scale . journal Water Research volume 248 , pages 120837

    work page 2024

  29. [29]

    , author Viviano, E

    author Nordio, R. , author Viviano, E. , author S \'a nchez-Zurano, A. , author Hern \'a ndez, J.G. , author Rodr \' guez-Miranda, E. , author Guzm \'a n, J.L. , author Aci \'e n, G. , year 2023 . title Influence of ph and dissolved oxygen control strategies on the performance of pilot-scale microalgae raceways using fertilizer or wastewater as the nutrie...

    work page 2023

  30. [30]

    , author Gil, J.D

    author Pataro, I.M.L. , author Gil, J.D. , author Guzm \'a n, J.L. , author Berenguel, M. , author Lemos, J.M. , year 2023 . title A learning-based model predictive strategy for pH control in raceway photobioreactors with freshwater and wastewater cultivation media . journal Control Engineering Practice volume 138 , pages 105619

    work page 2023

  31. [31]

    , author Sandoval, I.O

    author Petsagkourakis, P. , author Sandoval, I.O. , author Bradford, E. , author Zhang, D. , author del Rio-Chanona, E.A. , year 2020 . title Reinforcement learning for batch bioprocess optimization . journal Computers & Chemical Engineering volume 133 , pages 106649

    work page 2020

  32. [32]

    , author Radhakrishnan, T

    author Rajasekhar, N. , author Radhakrishnan, T. , author Samsudeen, N. , year 2025 . title Exploring reinforcement learning in process control: A comprehensive survey . journal International Journal of Systems Science , pages 1--30

    work page 2025

  33. [33]

    , author Morillas-Espa \ n a, A

    author Rodr \' guez-Torres, M.J. , author Morillas-Espa \ n a, A. , author Guzm \'a n, J.L. , author Aci \'e n, F.G. , year 2021 . title Modelling and pH control in raceway and thin-layer photobioreactors for wastewater treatment . journal Energies volume 14 , pages 1099

    work page 2021

  34. [34]

    , author Guzmán, J.L

    author Rodríguez Miranda, E. , author Guzmán, J.L. , author Acién, F.G. , author Berenguel, M. , year 2025 . title A comprehensive dynamic model of microalgae production in open raceway systems . https://doi.org/10.5281/zenodo.15579693

    work page doi:10.5281/zenodo.15579693 2025

  35. [35]

    , author del Rio-Chanona, E.A

    author Sachio, S. , author del Rio-Chanona, E.A. , author Petsagkourakis, P. , year 2021 . title Simultaneous process design and control optimization using reinforcement learning . journal IFAC-PapersOnLine volume 54 , pages 510--515

    work page 2021

  36. [36]

    , author Weisbarth, H

    author Schaum, A. , author Weisbarth, H. , author Meurer, T. , year 2017 . title Robust adaptive feedforward output-feedback tracking control for microalgae cultures . journal IFAC-PapersOnLine volume 50 , pages 12667--12672

    work page 2017

  37. [37]

    , author Eyckerman, R

    author Schepers, J. , author Eyckerman, R. , author Elmaz, F. , author Casteels, W. , author Latr \'e , S. , author Hellinckx, P. , year 2022 . title Autonomous building control using offline reinforcement learning , in: booktitle Advances on P2P, Parallel, Grid, Cloud and Internet Computing: Proceedings of the 16th International Conference on P2P, Parall...

    work page 2022

  38. [38]

    , author Kim, D

    author Seo, D. , author Kim, D. , author Son, H. , author Kim, Y. , year 2025 . title Implementation of reinforcement learning for enhanced pressure control in a 190,000-barrel crude distillation unit: The first full-scale commercial deployment . journal Engineering Applications of Artificial Intelligence volume 156 , pages 110907

    work page 2025

  39. [39]

    , author Gámez-Serrano, C

    author Sánchez-Zurano, A. , author Gámez-Serrano, C. , author Aci \'e n-Fern \'a ndez, F.G. , author Fern \'a ndez-Sevilla, J.M. , author Molina-Grima, E. , year 2021 . title Modeling of photosynthesis and respiration rate for microalgae--bacteria consortia . journal Biotechnology and Bioengineering volume 118 , pages 952--962

    work page 2021

  40. [40]

    , author Sowmya, G

    author Tarafdar, A. , author Sowmya, G. , author Yogeshwari, K. , author Rattu, G. , author Negi, T. , author Awasthi, M.K. , author Hoang, A. , author Sindhu, R. , author Sirohi, R. , year 2023 . title Environmental pollution mitigation through utilization of carbon dioxide by microalgae . journal Environmental Pollution volume 328 , pages 121623

    work page 2023

  41. [41]

    , author Kontoravdi, C

    author Wang, H. , author Kontoravdi, C. , author Del Rio Chanona, E.A. , year 2025 . title Offline reinforcement learning for bioprocess optimization with historical data , in: booktitle 14th IFAC Symposium on Dynamics and Control of Process Systems, including Biosystems (DYCOPS 2025): Slovakia, Bratislava, June 16-19, 2025

    work page 2025

  42. [42]

    , author Khoo, K.S

    author Wang, K. , author Khoo, K.S. , author Leong, H.Y. , author Nagarajan, D. , author Chew, K.W. , author Ting, H.Y. , author Selvarajoo, A. , author Chang, J.S. , author Show, P.L. , year 2022 . title How does the Internet of Things (IoT) help in microalgae biorefinery? journal Biotechnology Advances volume 54 , pages 107819

    work page 2022