Soft Sensing Based on Fuzzy Information Granulation

A kind of soft sensing model is proposed by combining methods of fuzzy information granulation (FIG) with support vector machine (SVM). The FIG has excellent performance of feature extraction and can reduce the dimension of the model sample in input data space. The extracted features are applied to SVM input for proceeding regressive modeling. Above measures are used to establish soft sensing model of diesel oil solidifying point. Results show that this model can get a good estimative accuracy and it is of wide extension as an effective method for data modeling.