Neuro-fuzzy model for energy prediction of different concrete dosages in buildings

Abstract Anincreasingprocess of urbanisation anda growing urban population heighten the need to understand the energy costsof the production of building materials. One of the most importanttoolsapplied to monitor the use of non-renewableenergyresources in the production of conventionalconcretes is energy input, intowhichfurther research is needed. In this study, an ANFIS (adaptive neuro-fuzzy inference system) hybrid model was developed to predict energy input in order to evaluate the energy demand required for each component of the production of conventional concrete (cement, water, fine aggregate and coarse aggregate) using 101 experimental dosages, 101 validation dosages and energy coefficients available in literature. The resultsshowedthatan adequate dosage can generate energy cost savingsof 24.77% in the production of concrete, while still maintaining the mechanical characteristics of compressive strength for conventional constructions.

摘要 快速推进的城市化进程与持续增长的城市人口，使得学界愈发需要深入认知建筑材料生产环节的能源成本。在普通混凝土（conventional concrete）的生产过程中，用于监测不可再生能源资源消耗的核心手段之一为能源投入量，针对该指标的深化研究仍有待开展。本研究构建了ANFIS（自适应神经模糊推理系统，adaptive neuro-fuzzy inference system）混合模型，以预测能源投入量，进而评估普通混凝土各生产组分——水泥、水、细骨料与粗骨料——所需的能源需求；研究共采用101组试验配合比、101组验证配合比，以及公开文献中的能源系数开展相关分析。研究结果表明，合理优化配合比可使混凝土生产的能源成本降低24.77%，同时仍可满足常规建筑工程对抗压强度等力学性能的要求。