Measurement method based on natural gamma ray activity and on X-ray backscattering combined with multivariate linear regression (MLR) calibration models have been shown to be useful for ash content determination in coal samples.
The performances of MLR models were compared with artificial neutral networks (ANN) models.
For selecting the best set of independent variables the stepwise backward elimination procedure was applied.
The standard error of calibration (SEC) and standard error of prediction (SEP) were calculated and used for the study of performance of the methods and calibration model. Gamma ray spectrometric measurements with multivariate linear calibration model were found to perform the prediction generally better than EDXRF method.
Both methods can be applied for the determination of ash content in coal with relative percent prediction error in the range 4-5%.