營建材料成本佔整體工程費用超過一半，有效控管物料費用能幫助承包商節省成本並創造更大的利潤，而鋼筋工程又為結構體工程中相當重要的一環，且鋼筋價格受整體鋼鐵業影響，其影響因子牽涉甚廣，導致承包商不易預測鋼筋價格之變化狀況，雖然現今有許多資料庫或預測模型能預估鋼鐵價格的消長，但大多屬於經濟取向的預測，較不符合承包商之需求。
隨著計算機領域中軟、硬體的快速發展，人工智慧中深度學習(Deep Learning)領域開始被廣泛研究，當中卷積神經網路(Convolutional Neural Network,CNN)能對於多維度的資料進行小範圍的特徵提取，使其在影像及語音辨識上有相當卓越的表現，也是促使近期深度學習成為相當熱門的原因之一。本研究將利用卷積神經網路處理圖像的特性應用在分析數值資料上。
為開發符合承包商需求之鋼筋價格預測模型，本研究首先以承包商角度解析模型需求，同時進行鋼筋價格因子分析，決定欲輸入模型之資料，接著以網路爬蟲程式自動化抓取所需數據，並建立訓練模型資料集。本研究為因應承包商需求，依欲預測之期數以及漲跌幅的區間建立不同的預測模型，並使用卷積神經網路演算法做為發展預測模型之基礎，於資料處理階段將數值資料轉化為圖片，透過卷積神經網路在圖像特徵提取的優異能力幫助訓練模型。在模型訓練完成後，執行其預測功能並根據預測結果模擬承包商之決策，輔助承包商在進行成本估算或擬定採購策略時，能準確預估鋼筋價格的變動，提升承包商在材料成本控管上的能力。

The cost of construction materials accounts for over 50% of the total construction cost. Hence, effectively monitoring the material price helps contractors to reduce cost and generate additional profit. Among the materials used in the construction phase, rebar is one of the most important ones utilized in the structure engineering. In addition, the price of rebar could be affected by many factors, which makes the prediction of rebar price a hard task for contractors. Although many prediction models are developed to estimate the price fluctuation, most of them are economy-oriented and are not suitable for contractor’s needs.
With the rapid development of the computer science, deep learning (a branch of artificial intelligence) has been widely studied. Convolutional Neural Networks (CNN), a class of deep learning, is capable of conducting small-range feature extraction within a multi-dimensional dataset, and thus has extraordinary performance in image and speech recognition. Therefore, this study aims to utilize the power of image processing of CNN on analyzing numerical dataset.
In order to develop a rebar price prediction model that meets the contractor’s needs, this study first analyzes the model requirements from the contractor’s perspective. In addition, factors that are related to rebar price are analyzed to determine input data for the model. Next, data is acquired by utilizing the automated web crawler program to build training datasets. Multiple CNN-based models are then built according to periods defined and fluctuation limits to meet the contractor’s needs. During the data processing stage, numerical data is converted to image and used to train the model with the help of CNN’s excellent ability in image feature extracting.
The result shows that the training and validation accuracy of models reach as high as 95% and 80%, respectively. Once the model has finished training, this study implements its prediction function and stimulate contractor’s decision based on the predicted result. This helps the contractors to estimate cost and set up purchase strategy by predicting the rebar price more precisely, and enhances the contractor’s ability on material price monitoring.

英文文獻
[1].A. G. Malanichev and P. V. Vorobyev, “Forecast of Global Steel Prices”, Studies on Russian Economic Development, Vol22: 304–311, 2011.
[2].Ashwin Bhandare, Maithili Bhide, Pranav Gokhale, and Rohan Chandavarkar, “Applications of Convolutional Neural Networks”, IJCSIT, vol7: 2206-2215, 2016
[3].FIP, http://www.idef.com, 1993.
[4].H Murat Günaydın, and S Zeynep Doğan, “A neural network approach for early cost estimation of structural systems of buildings”, International Journal of Project Management, Vol22: 595-602, 2004.
[5].Hao Wu, and Jinsong Zhao, “Deep convolutional neural network model based chemical process fault diagnosis”, Computers and Chemical Engineering, vol115: 185-197, 2018.
[6].Ismaail ElSawy, Hossam Hosny, and Mohammed Abdel Razek, “A Neural Network Model for Construction Projects Site Overhead Cost Estimating in Egypt”, IJCSI, Vol8: 273-283, 2011.
[7].Kaplan Andreas and Michael Haenlein, “Siri, Siri in my Hand, who's the Fairest in the Land? On the Interpretations, Illustrations and Implications of Artificial Intelligence”, Business Horizons, vol62:15-25, 2018.
[8].Mohammad Reza Moghaddam, Masood Manfezi, Amir Hossein Mehr Danesh, and Gholamhassan Kakha, “Prediction of monthly price of iron ore by using artificial neural network”, Indian Journal of Scientific Research, vol1:1200-1204, 2014.
[9].Neural Networks and Deep Learning, http://neuralnetworksanddeeplearning.com/ , Nielsen, Michael A , 2015.
[10].Preeti Kulkarni1, Shreenivas Londhe, and Makarand Deo “Artificial Neural Networks for Construction Management: A Review”, Soft Computing in civil engineering, vol1:70-88, 2017.
[11].Sumana Sharma, “An Integrated Knowledge Discovery and Data Mining Process Model”, Virginia Commonwealth University, 2008.
[12].Sumana Sharma, Kweku-Muata Osei-Bryson, and George M.Kasper, “Evaluation of an integrated Knowledge Discovery and Data Mining process model”, Expert Systems with Applications, vol39: 11335-11348, 2012.
[13].Trefor P.Williams, and JieGong, “Predicting construction cost overruns using text mining, numerical data and ensemble classifiers”, Automation in Construction Vol43: Pages 23-29, 2014.
[14].What’s the Difference Between Artificial Intelligence, Machine Learning, and Deep Learning? , https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/ , NVIDIA, 2016.
[15].Yun-Cheng Tsai, Jun-Hao Chen, and Jun-Jie Wang. “Predict Forex Trend via Convolutional Neural Networks”, Cornell University, 2018.
[16].Zhishuo Liu, Yongcong Wang, Shuang Zhu, Baopeng Zhang and Lingyun Wei, “Steel Prices Index Prediction in China Based on BP Neural Network”, LISS 2014, vol1: 603-608, 2015.

中文文獻
[1].CH.Tseng，初探卷積神經網路，ttps://chtseng.wordpress.com/2017/09/12/初探卷積神經網路/ ， 2017。
[2].呂思葦，應用知識本體論及BIM於施工架之佈設和風險模擬，碩士論文，國立成功大學土木工程學研究所，2017。
[3].李倩瑜，應用基因演算法結合時間序列於台灣地區鋼鐵價格漲跌幅之預測，碩士論文，國立臺北科技大學工業工程與管理系碩士班，2014。
[4].林大貴，深度學習人工智慧實務應用，博碩文化，新北市，2017。
[5].林秀貞，國際油價波動對重要營建材料成本影響之研究－以鋼筋、水泥、砂石、瀝青為例，碩士論文，國立中央大學土木工程研究所，2007。
[6].范光懿，營造業之成本控制實務，詹氏書局，台北，1996。
[7].許鈞甯，經濟指標與國際鋼鐵價格之關聯性－以中國、美國和日本為例，碩士論文，國立成功大學經營管理碩士學位學程， 2009。
[8].陳咏麟，國立大專院校鋼筋混凝土構造建築工程造價之預測與探討，第15屆營建工程與管理學術研討會，2011。
[9].陳玫英，製程工廠統包工程專案管理與物料管理之探討，碩士論文，國立台灣科技大學營建工程研究所，2002。
[10].陳建任，台灣鋼鐵產業發展趨勢與鋼價展望，金屬工業研究發展中心，2011。
[11].楊秉蒼，營建鋼筋損耗控實務，詹氏書局，台北，2007。
[12].詹智凱，鋼筋綁紮業經營現況與策略之研究，碩士論文，國立交通大學工學院工程技術與管理學程，2017。
[13].潘乃欣，林蔚菁，工程專案材料供應鏈模型之研究，科技學刊，第20卷，科技類，第1期，頁47-61，2011。