Predicting the heating value of rice husk with neural network

Abstract

Higher heating value (HHV) is an important property defining the energy content and thereby efficiency of fuels. In this paper, the correlation between the proximate analysis/ultimate analysis of some rice husk and HHV was discussed. It was found that the correlation between HHV and proximate, ultimate analysis was nonlinear. Therefore, two models were developed with artificial neural networks to predict HHV of rice husk by its proximate and ultimate analysis. A total of 25 samples of rice husk selected from the literature and experiments were used as the training data to build and train the two nets. Then, several samples selected randomly were used as predicting samples to check the accuracy of the two nets, respectively. A higher precision of 1.8% relative error in the prediction results was obtained through this method, while the relative error of linear empirical equations given in the literature was more than 12.7%. By this method, HHV can be estimated directly from the proximate analysis and ultimate analysis of rice husk when the HHV measurement equipment was not available.