碱激发体系与土聚水泥抗压强度关系的人工神经网络预测

摘要

The artificial neural network technology as a result of self-organization, self-learning, adaptive capacity, is often used in the gray system for model establishing, pattern recognition and target classification.This research uses MATLAB neural network toolbox,selecting the concentration of alkali activator ( COH-), the molar ratio of alkali-silica ( M20/SiO2 ) and the molar ratio of silicon (Al203/SiO2 ) as forecast indicators ,to predict the 28 d compressive strength of geopolymerie cement.After testing,this prediction reedel has higher prediction accuracy and can be used as production practice guidance.%人工神经网络技术由于自组织、自学习、自适应的能力,常被引入灰色系统模型建立、模式识别、目标分类等研究领域.本次研究运用人工神经网络技术,选取土聚水泥碱激发体系中的碱激发剂浓度(COH-)、碱硅摩尔比(M2O/SiO2)和硅摩尔比(Al2O3/SiO2)为预测指标,基于 MATLAB 神经网络工具箱,建立预测方法,对土聚水泥的 28 d抗压强度进行预测.结果表明,预测精度较高.