Leveraging Neural Networks to Optimize Heliostat Field Aiming Strategies in Concentrating Solar Power Tower Plants

TL;DR

This paper tackles the CSPT heliostat aiming problem by marrying data-driven constraint learning with neural network surrogates embedded in a mixed-integer optimization framework. A neural surrogate predicts a quality score that trades off energy capture against flux uniformity, and is reformulated as MILP via ReLU constraints to enable tractable optimization. An epsilon-constraint trust-region approach and progressive data sampling guide iterative refinement, improving flux uniformity while controlling hotspots. In a Dunhuang CSPT case study, the proposed NN+Opt method yields flatter flux distributions and safer thermal conditions with only a modest reduction in total energy, outperforming traditional sweep heuristics. The work demonstrates the value of integrating learning and optimization for complex, nonlinear solar-thermal systems and suggests extensions to dynamic and robust planning."

Abstract

Concentrating Solar Power Tower (CSPT) plants rely on heliostat fields to focus sunlight onto a central receiver. Although simple aiming strategies, such as directing all heliostats to the receivers equator, can maximize energy collection, they often result in uneven flux distributions that lead to hotspots, thermal stresses, and reduced receiver lifetimes. This paper presents a novel, data-driven approach that integrates constraint learning, neural network-based surrogates, and mathematical optimization to overcome these challenges. The methodology learns complex heliostat-to-receiver flux interactions from simulation data, constructing a surrogate model that is embedded into a tractable optimization framework. By maximizing a tailored quality score that balances energy collection and flux uniformity, the approach yields smoothly distributed flux profiles and mitigates excessive thermal peaks. An iterative refinement process, guided by the trust region and progressive data sampling, ensures the surrogate model improves the obtained solution by exploring new spaces during the iterations. Results from a real CSPT case study demonstrate that the proposed approach surpasses conventional heuristic methods, offering flatter flux distributions and safer thermal conditions without a substantial loss in overall energy capture.

Figures (11)

• Figure 1: Schematic representation of a CSPT plant.
• Figure 2: Heliostat field layout at Dunhuang 10 MW$_e$ plant. Rows are colored as a function of distance to the receiver for the sake of heliostat identification in the aiming maps.
• Figure 3: Receiver divided into $P$ panels, 2D representation of the panels in the receiver, and schematic representation of vertical mean concentration calculation.
• Figure 4: Schematic representation of aiming strategy quality score calculation: (a) Area calculation under the vertical mean concentration curve. (b) Penalty for deviation from uniform shape in the central part of the panel.
• Figure 5: NN structure for optimizing heliostat field aiming strategies.