In order to decrease the measurement error of the traditional dual-energy γ-ray method and reduce the impact of the ash component on detection accuracy in ash determination, a new coal ash measurement model based on fuzzy neural network and the dual-energy γ-ray was established, and then the model was applied on online testing of coal ash content. The application of the traditional and new coal ash measurement model was analyzed by online testing, and the measurement results of both were comparatively analyzed. The results show that the relative error of ash content using the new method was less than 1%, while the average relative error of the traditional measurement was about 3%, and the measurement accuracy of the new method is not affected by the ash component. Meanwhile, the relationship between the chemical composition of coal ash and ash content was analyzed by using X-ray fluorescence spectrometry (XRF), and the results indicate that the content of Fe, Ca, Mg and S elements in coal can affect the detection accuracy of traditional dual-energy γ-ray dialysis, the greater fluctuations of their ingredient content result in greater measurement error.