Akademik Veri Yönetim Sistemi
Smart Agricultural Technology, cilt.12, 2025 (Scopus)
Achieving precise seed distribution in pneumatic precision planters is essential for optimizing crop yield and resource efficiency. This study introduces an AI-assisted multi-objective optimization approach, where ChatGPT played a central role in optimizing seeding performance by refining disk speed, vacuum pressure, and area of the vacuum hole on the disk. By leveraging ChatGPT for method selection, parameter tuning, and result analysis, the optimization process was streamlined, reducing reliance on traditional trial-and-error approaches and enhancing decision-making efficiency. Two distinct optimization approaches were employed: (i) Multi-Criteria Decision-Making (MCDM) methods—MOORA and TOPSIS—to rank existing parameter sets, and (ii) NSGA-II, an evolutionary algorithm that generated new Pareto-optimal solutions by balancing multiple performance criteria. Optimization was conducted based on laboratory experiments using onion and sugar beet seeds, with performance evaluated through quality of feed index (QFI), multiple index (MULTI), miss index (MISS), and preciseness (PREC). ChatGPT, as an advanced AI tool, facilitated method selection, implementation, parameter tuning, and result interpretation. MOORA & TOPSIS provided deterministic optimal parameter combinations, identifying 0.15 m/s disk speed, 1 kPa vacuum pressure, and 3.1 mm² hole area for onion seeds, and 0.2 m/s disk speed, 1.75 kPa vacuum pressure, and 4.7 mm² hole area for sugar beet seeds. In contrast, NSGA-II generated a Pareto-optimal solution set, offering a flexible trade-off between QFI maximization and error minimization. To ensure reliability, NSGA-II results were validated through experimental testing, confirming that the algorithm's predicted optimal settings were within ±5 % of actual measured values. The findings underscore the potential of AI-assisted methodologies, particularly ChatGPT-driven optimization, in agricultural mechanization, enhancing efficiency and reducing dependency on labor-intensive trials.