Rose pattern for heliostat field optimization with a dynamic speciation‐based mutation differential evolution

Abstract

Concentrated solar power technology is one of the most promising technologies in energy field. Arguably, the heliostat field layout is a crucial component in solar power tower system. Numerous studies have been developed for heliostat field optimization. However, most of the existing layouts which utilize radial staggered patterns are based on only two or four variables, leading to relatively rigid modes due to strong configuration constraints. In this article, we propose a new method called rose layout, which divides the regular radial staggered pattern into six sectors and they are optimized separately. Therefore, the radial increments between consecutive rows are not restricted to zones or rows, only relevant to which sector they belong to. This arrangement is more flexible and also efficient. Furthermore, a new differential evolution algorithm with a dynamic speciation-based mutation strategy (DSM-DE) is developed to solve this high-dimensional problem. In order to validate the proposed rose pattern and DSM-DE, three sets of comparative experiments were carried out. The first set of tests were operated with the conventional four optimization variables, the second series optimized total 43 radial increments between consecutive rows, whereas the third series employed the rose layout. All sets of cases were optimized by four competitive variants of differential evolution algorithm, ie, JADE, SHADE, EB-LSHADE, and DSM-DE. Experimental results verified that the rose layout can obtain higher overall optical efficiency and less land coverage than previous methods and DSM-DE is superior to other DE variants for this high-dimensional problem. The heliostat field studied in this article is simulated in Qinghai, China. By integrating rose layout with DSM-DE, the field unweighted efficiency progressed from 44.386% to 53.972%, and the annual weighted efficiency reached 59.091%, which was 0.318% higher than the 43-variable optimization.