演化式人工智慧建立專案實獲完工工期推論模式之研究

Evolutionary Artificial Intelligence Based Earned Schedule Inference Model for Construction Project

施工過程中，受限於環境、天氣等眾多因素影響，造成完工工期經常難以準確掌控，施工單位在預測工期時，必須仰賴過去之經驗，無法即時反映影響工期之因素並利用工程現況客觀地預測完工工期。本研究以建築工程之建築物完工工期為研究標的，不包含機電、裝修工程等工期，應用演化式人工智慧作為推論模式之核心SOS-LSSVM(Symbiotic Organisms Search-Least Squares Support Vector Machine)，透過案例學習發展建立專案實獲完工工期推論模式找出每期輸入變數與待完工成本之間的映射關係，進而計算預估完工工期(Estimate Schedule At Completion, ESAC)。藉此作為施工過程中作為時程管控的參考依據，以達到提前預警的目的。結果顯示本研究發展之模式RMSE低於0.03、MAPE低於10%、MAE低於3%及相關係數達0.99，相較傳統實獲值工期預測及其他人工智慧模式具較佳預測準確率。經實際案例分析對於管理者能夠有效進行時程之管控且降低工程成本。

 

Because of factors such as the environment and weather during a construction process, accurate control of schedule at completion (SAC) is often difficult. Builders must rely on past experience to predict the project duration. Thus, they often cannot react punctually to factors affecting the construction duration or predict objectively the SAC by using the project’s current progress. This study developed an SAC inference model for building structure using the basis evolutionary artificial intelligence - Symbiotic Organisms Search-Least Squares Support Vector Machine (SOS-LSSVM). Through training with these historical cases, it was used to map the relationships between the input variables and the cost of construction work to be completed.The learning results indicated good performance, with Root Mean Square Error (RMSE) of less than 0.03, a Mean Absolute Percentage Error (MAPE) of less than 10% , d a Mean Absolute Error (MAE) of less than 3% and a correlation of 0.99, proving the SOS-LSSVM model as more reliable than the currently prevailing method. In case study, the proposed model for provides more accurate results for assisting managers with schedule and cost management.