Aiming at the online testing problem of outlet concentration in the evaporation process of alumina production, a
multi-support vector machine modeling method based on the gray relational analysis and kernel fuzzy ??-means clustering in
the evaporation process is proposed. Variable factors of greater correlation with outlet concentration are selected by using
grey relational analysis, these training samples constituted by these factors are divided into subsets with different cluster
center by using kernel fuzzy ??-means clustering, and output is obtained finally by summing each support vector machines
model’s output comprehensive weighting. The experimental results of industrial production data of evaporation process
show that, compared with kernel fuzzy cluster method and least squares support vector machine(KFCM-LSSVM) model, the
proposed model can get good robustness, high accuracy, and better generalization, which can be used for the optimal control
of the evaporation process.