Atomization cleaning rate research for mold boxes based on Adaptive Neural Fuzzy Inference System

in the extant condition, it was vital to clarify the influence from known specific factors to cleaning efficiency of mold box in the starter-maker machine. The Fen liquor holds the extremely important status in China, but massive water was wasted in the exiting cleaning process. Adaptive Neural Fuzzy Inference System (ANFIS), intelligence computing system based on fuzzy mathematics and neural network, had been built in this paper. 225 groups of experiments data were used in training and checking processes separately for the ANFIS based on two different membership functions, Triangle and Gaussian. After the contrast, it was discovered that the choice of membership function could influence the forecast precision extremely. One system, whose forecast average error of cleaning rate was merely 0.87%, and the cleaning rate forecast surfaces, helping designer to understand the key factors influencing the cleaning efficiency, were obtained. The multi-objectives optimization research could be conducted based on this accurate forecasting system. Using ANFIS is an effective method for simulating the processes whose internal action mechanisms have not been realized clearly.