Abstract: Determination of ash of flotation concentrate is of vital importance to automated find coal flotation process.However,the traditional method of rapid determination of ash through burning cannot meet the requirement of such a process.A flotation concentrate ash prediction model based on foam image is,therefore,designed through first denoising processing and then sieving of the characteristic factors affecting the concentrate ash using MIV value evaluation method.Following these processes,a RBF neural network-based ash prediction model is finally developed.As evidenced by comparison with the BP neutral network-based method,the RBF network-based model can produce a far better concentrate ash prediction result.