Cost-effective economic dispatch in large-scale power systems using enhanced manta ray foraging optimization

Abstract Economic dispatch (ED) is a critical optimization problem in power systems, challenged by real-world constraints such as prohibited operating zones (POZ), valve-point loading effects (VPL), and multi-fuel options (MFO). POZ are regions where generators cannot operate due to mechanical limitations; VPL introduces cost fluctuations caused by turbine valve operations; and MFO allows generators to switch between multiple fuel types, adding complexity to cost functions. To address these challenges, this study proposes four enhanced Manta Ray Foraging Optimization (MRFO) variants: opposition-based MRFO (OMRFO), quasi-oppositional MRFO (QMRFO), opposition-based generation jumping MRFO (JOMRFO), and quasi-oppositional generation jumping MRFO (JQMRFO). These variants leverage opposition-based learning (OBL), quasi-oppositional learning (QOBL), and a generation jumping mechanism to balance exploration and exploitation, overcoming limitations of standard MRFO, such as slow convergence and local optima entrapment. OBL and QOBL diversify the search by generating opposite or quasi-opposite solutions, expanding the search space, and avoiding stagnation. The jumping mechanism introduces probabilistic "jumps" to explore non-adjacent regions, enhancing exploration further. Exploitation is refined by retaining and improving the most promising solutions. The algorithms are tested on standard benchmark systems widely used in power systems literature, including the 10, 15, 40, 140, and 160-unit systems, ensuring comparability and reproducibility. Results show that JOMRFO outperforms MRFO and other state-of-the-art methods, achieving significant annual cost savings: $3,730,344 on a 10-unit system, $54,641,376 on a 40-unit system, and $955,501,418 on a 140-unit system. These findings highlight the effectiveness of the proposed variants in improving optimization efficiency and reducing operational costs.