Soft sensing modeling based on dynamic fuzzy neural network for penicillin fermentation

Since the establishment of initial model and determination of the rules´ number of traditional fuzzy neural network both rely on experiential knowledge, so a soft sensor modeling method using dynamic fuzzy neural network is proposed in this paper. The network structure is based on extended radial basis function neural network. Sequential learning method is utilized for parameter estimation and structure identification, and then the pruning technique is introduced to make the structure more compact. The network structure identification is equivalent to determination of fuzzy rules. And the auxiliary variables are identified by uniform incidence degree algorithm. Take the key biological parameters soft sensing of penicillin fermentation process as an example, the simulation results show that this proposed method has satisfied modeling precision and practicality.