The effects of calcium nitrite solution, fly ash, silica fume, and their co
mbination on the alkali-silica reaction were studied experimentally accordi
ng to ASTM C 227 (mortar bar test). To explore the mechanism of such effect
s, chemically combined water tests and X-ray diffraction examinations were
also conducted. The microstructural information provided by these experimen
ts was in agreement with the macroscopic observations. A neural network ana
lysis was then conducted to study the feasibility of the prediction of a th
eoretical model. To solve the consequent nonlinear coupled equations, Some
globally convergent skills were adopted. The results showed that a cascade-
correlation algorithm could produce reasonable outputs that coincide with t
he experimental results after proper training.