Improvement of the empirical regression type algorithm, based on the well-known canonical form equation by Lachance, for calculations of chemical element concentration and its applications for XRF expressanalysis to control and manage the technological processes on mine manufactures, are considered.

Despite the criticism of regression-type algorithms, because they have no physical sense, they are used successfully in the software of xray quantometers and energy dispersive spectrometers at different plants when it is necessary to analyze the samples with slightly varied matrix for the concentration of the elements of interest very rapidly (2-3 minutes for full analysis of one sample) and with high accuracy (-1% relative error in weight).

Our task was to improve the so-called physics of the algorithm to get the required analytical stability and to account for the random factors which influence measured intensities unexpectedly (heterogeneity of analyzed sample, roughness of its surface, etc.).

The results of the investigations were applied to the samples of apatit-nephelin manufacture. Together with successful solution of the problem, new analytical possibilities for classification of samples according to their mineralogical differences were obtained.

The new XRF empirical regression type algorithm was realized in special software.