Machine learning-based hybrid dynamic modeling and economic predictive control of carbon capture process for ship decarbonization

TL;DR

This work tackles shipboard decarbonization by integrating a solvent-based post-combustion carbon capture plant with the ship’s engine system and heat recovery. It introduces a learning-enabled hybrid dynamic model that fuses imperfect first-principles physics with two neural networks to infer algebraic states and compensate state dynamics, enabling a robust economic model predictive controller (EMPC). The EMPC, solved efficiently via a cross-entropy method, minimizes energy use and $CO_2$ emissions under varying engine-load conditions, demonstrating improved data efficiency and generalization relative to purely data-driven models, and achieving notable economic savings over conventional MPC. The approach offers practical potential for energy- and cost-efficient shipboard PCC operation and provides a framework for integrating physics with data-driven corrections in complex, large-scale industrial processes.

Abstract

Implementing carbon capture technology on-board ships holds promise as a solution to facilitate the reduction of carbon intensity in international shipping, as mandated by the International Maritime Organization. In this work, we address the energy-efficient operation of shipboard carbon capture processes by proposing a hybrid modeling-based economic predictive control scheme. Specifically, we consider a comprehensive shipboard carbon capture process that encompasses the ship engine system and the shipboard post-combustion carbon capture plant. To accurately and robustly characterize the dynamic behaviors of this shipboard plant, we develop a hybrid dynamic process model that integrates available imperfect physical knowledge with neural networks trained using process operation data. An economic model predictive control approach is proposed based on the hybrid model to ensure carbon capture efficiency while minimizing energy consumption required for the carbon capture process operation. The cross-entropy method is employed to efficiently solve the complex non-convex optimization problem associated with the proposed hybrid model-based economic model predictive control method. Extensive simulations, analyses, and comparisons are conducted to verify the effectiveness and illustrate the superiority of the proposed framework.

Figures (21)

• Figure 1: A schematic diagram of the entire plant consisting of the ship engine system and the shipboard post-combustion carbon capture process. The black lines represent mass flow streams, and the red lines represent energy flow streams. $x$, $z$, $u$, $p$, and $y$ denote the differential states, algebraic states, control inputs, known disturbance, and controlled outputs of the entire plant.
• Figure 2: An illustrative diagram of the proposed hybrid model that comprises imperfect first-principles knowledge and dense neural networks.
• Figure 3: A block diagram illustrating the training process of the hybrid model.
• Figure 4: A graphical illustration of the recursively updating process of the cross-entropy method.
• Figure 5: Trajectories of the control inputs and the known disturbance generated for modeling.

Theorems & Definitions (3)

• Remark 1
• Remark 2
• Remark 3