Real-time in situ monitoring of ash content in coal via dual-energy X-rays and an MLP neural network

Abstract

ABSTRACT The accurate estimation and measurement of coal ash are crucial for fuel selection, combustion efficiency assessment and quality control. However, the most widely used ash measurement method is combustion; this method is highly accurate but has a certain lag; additionally, an older radiation measurement method has a certain error. In this regard, in this study, a detection model and measurement method are proposed based on the combination of dual-energy X-ray measurement results and artificial intelligence algorithms, i.e. a coal ash detection model based on a multilayer perceptron (MLP) neural network. A model training database was created, and 6468 raw data points were measured with an experimental apparatus and organized. The results showed that the root-mean-square error between the predicted value and the true value of the trained model was 0.0857. By comparing several indices with the traditional backpropagation (BP) neural network, the root-mean-square error was reduced by 1.27%, and the model’s errors in different predicted output values were uniformly distributed without evident systematic deviation; these results demonstrated and confirmed that our proposed ash prediction model achieved high estimation accuracy and had strong robustness.