資料驅動最佳化之產業應用

Abstract

在數位轉型時代，各行各業正利用先進的數據驅動優化技術來提高效率、降低成本並改善決策過程。這些技術結合了大數據、機器學習和人工智慧，通過創建模型和模擬來預測結果並提供最佳策略，徹底改變了傳統做法。然而，現實世界中的優化問題往往是複雜的、多目標的且資源密集型的，需要使用進化算法和基於模擬的優化等複雜的方法。替代模型等技術有助於減少計算成本，但會引入近似誤差。製造業、能源、海事和農業等行業從數據驅動優化中受益顯著，應對數據稀缺、噪聲和不平衡等挑戰。這些技術的框架包括數據收集、模型開發和計算，確保模型的穩健性和適應性。本研究探討了在製造排隊時間環生產系統、海洋渦輪模擬校準和海洋農場佈局優化中的應用，展示了性能和效率方面的顯著改進.

In the digital transformation era, industries are leveraging advanced data-driven optimization techniques to enhance efficiency, reduce costs, and improve decision-making processes. These techniques integrate big data, machine learning, and artificial intelligence, revolutionizing traditional practices by creating models and simulations to predict outcomes and suggest optimal strategies. However, real-world optimization problems are often complex, multi-objective, and resource-intensive, requiring sophisticated approaches like evolutionary algorithms and simulation-based optimization. Techniques such as surrogate models help mitigate computational costs but introduce approximation errors. Industries like manufacturing, energy, maritime, and agriculture benefit significantly from data-driven optimization, addressing challenges like data scarcity, noise, and imbalance. A framework for these techniques involves data collection, model development, and computation, ensuring robust and adaptable models. This study explores applications in manufacturing queue time loop production systems, marine turbine simulation calibration, and marine farm layout optimization, demonstrating significant improvements in performance and efficiency.