Bioprocess Modeling Using Fuzzy Regression Clustering and Genetic Programming

Abstract

In an industrial Avermection bioprocess, there are some different phases and some interims, so using fuzzy clustering technique to partition the whole process and using several models to represent these different phases respectively is more reasonable. Here, we built a fuzzy model for the bioprocess by a mixture method which integrates fuzzy regression clustering technique, genetic programming (GP), genetic algorithm (GA) and interpolation technique. Fuzzy regression clustering technique is used to partition the whole input space into several subspaces based on whether the training data having a similar model which is identified by GP, GA is used to optimizes the parameters of these models and interpolation technique used to define the membership grade for the input data. By this approach, we can fuzzily partition the whole process, find the structures of models which represent subspaces respectively and estimate the parameters simultaneously. Moreover, it has more chance to get a solution with better generalization.