工程進度規劃自動評核與建議系統架構之研究

Abstract

工程進度規劃是營建工程專案執行順利與否的重要關鍵之一。隨著工程專案逐漸地巨型化與複雜化，進度工程師所規劃的進度網圖難免會有筆誤、矛盾甚至不合法規的錯誤發生。由於進度網圖錯綜複雜，這些錯誤往往卻難以透過人力去核對確認。本研究透過文獻回顧、專家訪談與實際觀察等三種知識擷取方式，對高層建築工程進度規劃的相關專家知識作完整的歸納與分析，並依其原因與目的、立場觀點以及評核型態等三方面分類探討之。有別於過去的知識庫建立方式，本研究建立排程評核法則之一般化敘述模式，以物件導向形式紀錄專家知識。此外，系統使用者亦可自行建立符合個人要求與需求的排程評核法則。本研究亦依據所建立之評核法則一般化模式，建立電腦自動評核系統 - ScheduleCoach。該系統以法則式推理的方式對工程進度規劃各部分的屬性資料作自動評核。ScheduleCoach 系統對於不符合評核法則的作業項目、作業關係或使用資源，引用案例式推理的方式經過檢驗與修正之後提出排程修正建議，供進度規劃者作為修改的參考。

Planning and scheduling is one of the key factors to the success of a construction project. Because of the increasing complexity and size of construction projects nowadays, pre-construction schedules tend to comprise a large number of activities and much related information. These schedules often contain unintentional errors and conflicts, and those due to the lack of experience. These mistakes need to be corrected, and often they can be corrected easily by an experience scheduler. However, due to the number of activities involved, sometimes these mistakes are hard to be found.With the increasing power and decreasing cost, personal computers have become a common tool used by construction experts to facilitate their daily decisions such as scheduling, cost estimation, and cash flow analysis. This research applies artificial intelligence techniques (i.e., case-based reasoning and rule-based reasoning) and develops a computer system called ScheduleCoach, which can analyze a computerized schedule and provide corrective advises. The current knowledge base of the system focuses on high-rise building construction.