以多目標最佳化演算法找尋最適融資辦法之施工排程—以台灣營建工程為例

Abstract

營造廠在財務特性上有自有資金偏低及偏好短期融資等特性，而營造廠商通常以建築融資做為興建期間取得營運資金的方式之一，如何在專案生命週期的前端，準確地估計計畫成本，找出適合的融資辦法，為專案生命週期前端的重要課題之一。 本研究以台灣營造業為例，將營建融資加入專案現金流計算，應用FPGA（Fast Pareto Genetic Algorithm）找尋總工期最小化、融資成本最小化、需自備之自有資金最小化以及總利潤最大化多目標下最適融資辦法之排程，並以窮舉法（Enumeration）、粒子群優化法與模擬退火演算法來驗證本系統之可信度、可靠度與實用性。 結果顯示本系統找到之最適解相當位於全域柏拉圖解之前20%，並可減少近達99.99%之花費時間。本研究建立之系統可協助營造廠商於專案生命週期前期找出最適之融資辦法之排程以及其現金流，增進營造廠商之財務管理及專案評估與預測。

For construction companies in Taiwan, low own funds and short-term financing are two of the financial characteristics, and financing from banks is a commonly-used way to obtain working capital during construction stage. To accurately estimate project costs and conduct appropriate project financing plans are considered two important topics in the beginning stage of project’s life cycle. In this research, case study of Taiwanese construction project is analyzed by FPGA（Fast Pareto Genetic Algorithm）, which combining construction financing and working capital into calculation to define an optimal construction schedule based on four objectives: Minimized total duration, minimized financing cost, minimized own funds, and maximized total profit. Verification is done by applying Enumeration method, Particle Swarm Optimization（PSO）, and Simulated Annealing Arithmetic（SAA） to verify the credibility and reliability, and practicality of this proposed system. The results show that the optimal solution is equivalently to the top 20% located in the Pareto solutions, and FPGA can reduce nearly 99.99% of the total time spent. The system proposed in this study can assist construction companies in finding an optimal construction schedule with compatible financing plans and cash flows in the early project life cycle, and also facilitating the operation of construction financial plan and increasing efficiency in project management.