运用模糊神经网络研究粉煤灰抑制AAR问题

提出运用模糊神经网络研究粉煤灰抑制AAR问题。利用基于试验数据运算得到的模糊神经网络获得整体(系统)意义上的模糊规则，再利用该训练好的网络扩充分析所需的其它模糊规则。通过对所有规则进行全面、深入的分析，找出粉煤灰抑制AAR的整体规律。实例计算表明，引起AAR膨胀的主要因素是胶凝材料中水泥碱量所占比例，其次为胶凝材料中粉煤灰掺量和胶凝材料中粉煤灰碱量所占比例。水泥碱量越大，AAR膨胀越大；粉煤灰掺量越大，AAR膨胀越小；粉煤灰碱量越大，AAR膨胀越大。

The approach of applying fuzzy neural network in research on the problem of fly ash in suppressing expansion due to AAR

The approach of applying fuzzy neural network ( FNN) in research on the problem of fly ash in suppressing expansion due to alkali-aggregate reaction (AAR) is proposed for the first time. The fuzzy rules of obeying whole law are obtained by use of the FNN which is trained by inputting all of the test data in the FNN, and then other fuzzy rules needed for study are complemented using the trained FNN. By studying all the rules obtained comprehensively the whole law of fly ash in suppressing expansion due to AAR is discovered. The calculation shows that it is the alkali content of cement in cementitious material that plays a primary role on expansion due to AAR while the lever of fly ash-replacement and the alkali content of fly ash plays secondary role on its effect of suppressing expansion due to AAR. The more the alkali content of cement in cementitious material, the larger the AAR expansion is. The more the lever of fly ash-replacement, the less the AAR expansion is. The more the alkali content of fly ash, the larger the AAR expansion is.