營建工程具備高度不確定性與複雜的風險特性，現行實務界普遍採用 的風險矩陣常面臨多項危害同分而難以區分控管優先順序之困境。雖然模 糊理論已被引入風險評估領域以彌補量化不足，但現有研究多傾向對所有 工項套用單一隸屬函數設定，忽略了不同分項作業間資訊明確度的本質差 異。本研究旨在建立一套結合三角模糊數（TFN）與梯形模糊數（TrFN）之 「混合型模糊風險評估模型」，透過差異化之隸屬函數設定策略，提升 Fuzzy TOPSIS 演算法在營建風險排序上的鑑別力與實務合理性。 研究過程首先透過參數調整模擬，探討語意變數之分布間隔與不對稱 性對排序結果的影響機制；隨後針對實際營造專案中「基礎工程」、「結構 體工程」與「外牆裝修工程」之 64 項潛在危害進行數值驗證。研究結果 顯示，混合型評估模型能有效捕捉專家在面對高不確定性工項時的真實判 斷心理，顯著改善了傳統模型在關鍵風險排序上的模糊性。此外，本研究 將複雜的模糊運算邏輯整合於自動化運算模型中，確保管理人員僅需輸入 直覺之語意評分即可獲得具備科學依據的精確排名。本研究不僅驗證了差 異化隸屬函數設定的理論優勢，亦提供營建產業一套兼顧精準度與易用性 的風險決策升級工具。;Construction projects are characterized by high complexity and inherent uncertainties. Traditional risk matrices， widely used in the industry， leading to frequent "ties" in risk rankings that hinder effective resource allocation and decision-making. While fuzzy logic has been introduced to address these quantitative deficiencies， existing research typically applies a uniform membership function across all tasks， neglecting the significant variations in information certainty inherent in different sub-projects. This study develops a Hybrid Fuzzy Risk Assessment Model that integrates Triangular Fuzzy Numbers (TFN) and Trapezoidal Fuzzy Numbers (TrFN) within a Fuzzy TOPSIS framework. The core objective is to enhance the discriminative power and practical rationality of risk rankings through differential membership function strategies. The methodology first involves parameter adjustment simulations to investigate how the interval and asymmetry of linguistic variables influence ranking outcomes. Subsequently， the model is validated through a case study involving 64 potential hazards across foundation， structural， and finishing works. The results demonstrate that the hybrid model effectively captures the psychological judgment of experts in high-uncertainty scenarios and significantly improves the precision of critical risk rankings compared to traditional uniform models. Furthermore， the complex fuzzy algorithms are encapsulated within an automated Excel-based module， ensuring that practitioners can generate scientifically grounded rankings simply by entering intuitive linguistic ratings. This research not only validates the theoretical advantages of differential membership function settings but also provides the construction industry with a decision-support tool that balances precision with ease of use.
