Abstract：Under the same lighting conditions, the MATLAB software was used to control the industrial camera in order to obtain the image information of the different ignition loss for fly ash. Taking into account the light reflectance differences of different components in the fly ash, extraction of different components of the triboelectrostatic beneficiation fly ash Image feature parameters, the extreme learning machine neural network was used to establish a mathematical model between the ignition loss and image characteristics, and the best activation function was got by comparing the prediction effect of loss on ignition, and the on-line rapid detection of the ignition loss was realized. The results show that the prediction model established by extreme learning machine can accurately identify the image characteristics, and quickly obtain the the ignition loss of fly ash. The extreme learning machine is high precision, and it provides a technical reference for the rapid online testing of the fly ash ignition loss in industrial production.

Keywords：fly ash; loss on ignition; image characteristics; mathematical mode; extreme learning machine