Soft sensing system for coal storage in ball mill based on fuzzy neural network

A new on-line measurement approach for coal storage in ball mill is presented in this paper. In the approach, fuzzy neural network modeling is adopted to construct the measurement model for coal storage in ball mill. During the modeling, fuzzy neural network model is employed to approximate the non-linearity of coal storage in ball mill. Speed of coal speed, sirocco flux, recycles wind flux, export temperature and pressure difference are selected as the measure variables. Subtractive-clustering algorithm is used to determine the optimum number of clusters; this made the fuzzy neural network model simple and accurate. Based on it, an instrument based on PC104 is developed and the test in laboratory is conducted. The test result shows that the measure error is lower and acceptable. It lays a foundation for the optimal control of coal storage in ball mill.