Use of artificial neural network for modelling the interaction between fly ash and lime under steam curing

The chemical interaction between fly ash and lime in steam cured fly ash-lime compacts was modelled by artificial neural network to predict the free lime remaining in the mixes after the curing period. Process parameters, like, the pozzolanicity of the ash samples, their surface areas, unburnt carbon content, curing period and the proportion of lime in the fly ash-lime mixes were taken as the inputs for the model and the free lime remaining in the mix was taken as the output parameter. A generalized feed forward back propagation three layered neural network model was used with tan hyperbolic transfer function at both the input and the output layers with 400 exemplars. For the training data after 3000 iterations the mean square error (MSE) value was found to be the minimum for the prediction mode. The model when tested for the test data the difference between the actual value of the free lime and the predicted value of the free lime content in the mixes after different periods of curing was found to be within +/-10%.