Abstract: Online detection of mineral elements contained in coal provides a crucial data support for realizing control of ash of products with feedback ash data. Following an introduction to and comparison of the XRF, LIBS and PGNAA techniques for online detection of mineral constituents contained in coal, the paper goes to elaborate on the working principle, make-up and structural design of the PGNAA online muti-element detection technique and particularly the effects obtained in field application of this technique in mine and central coal preparation plants, as well as the comparison of the actually measured and analysis data. Practice shows that under varying operating conditions and quality requirements of products, compared with the analytical data, the standard deviation of the ash values actually measured with the PGNAA technique can be controlled within 0.15%, with a correlation coefficient being over 0.91; and for the detection of total sulfur and Fe₂O₃, the standard deviations and correlation coefficients are 0.032% and 0.995, 0.025% and 0.950, respectively. It can be seen that the actually measured ash, total sulfur and Fe₂O₃ data are in good agreement with the analytical data. With the ongoing development of intelligent coal preparation plants, an ever stricter requirement is placed online coal analysis units. Looking ahead, the high-accuracy and high-adaptability PGNAA technology will play an ever bigger role in raw coal homogenization, separation control, accurate coal blending and rapid determination of quality of delivered commercial coal.