Researchers from the University of Chinese Academy of Sciences, including Mingxue Yan, Xuewen Zhang, Kaixiang Zhang, Zhaojian Li, and Xunyuan Yin, have developed a new method for optimizing the economic performance of nonlinear systems, with potential applications in the energy sector. Their work was recently published in the journal Automatica.
The team proposes a data-enabled economic predictive control method designed to optimize the operational performance of nonlinear systems while respecting hard constraints on system outputs. To achieve this, they use neural networks to create two lifting functions that map inputs and outputs into a higher-dimensional space. In this space, the nonlinear economic cost function can be approximated using a quadratic function, making it easier to handle mathematically.
The researchers extend their predictive control framework to address nonlinear dynamics by using the mapped inputs and outputs, which form a virtual linear representation of the original nonlinear system. They also reconstruct the system output variables from the mapped outputs, allowing them to impose hard output constraints. By formulating the online control problem as a convex optimization problem, they ensure that it can be solved efficiently, despite the nonlinearity of the system dynamics and the original economic cost function.
The effectiveness of the proposed method was demonstrated through two large-scale industrial case studies: a biological water treatment process and a solvent-based shipboard post-combustion carbon capture process. These examples highlight the potential of the method to improve the economic performance of energy-related processes while respecting important constraints.
For the energy industry, this research offers a promising approach to optimizing the operation of complex, nonlinear systems. By enabling more efficient and cost-effective operation, the method could contribute to the development of more sustainable and economically viable energy solutions. The research was published in the journal Automatica, a leading publication in the field of control engineering.
This article is based on research available at arXiv.