Abstract:To solve the online detection difficulty of the key state variables in fermentation process with instruments, a soft sensing modeling method which was based on continuous hidden Markov model (CHMM) was proposed. In order to improve the robustness of CHMM, multi-observation training sample sequences were adopted to train the CHMM, and the modified BaumWelch parameters revaluation formula was used to optimize the parameters of CHMM. The new observation vector was inputed into the CHMM model library, and the emission probability of each CHMM in the model library was calculated by Viterbi algorithm. The soft sensing result was obtained by computing the weighted average. The proposed modeling method was applied to the soft sensing modeling of cell concentration in the erythromycin fermentation process. The modeling and simulation were also complished. The results show that the CHMM soft sensing model has high prediction accuracy of cell concentration for fermentation process, which is better than that of artificial neural networks soft sensing model.
刘国海, 江兴科, 梅丛立. 基于连续隐Markov模型的发酵过程关键状态变量软测量[J]. 江苏大学学报(自然科学版), 2011, 32(4): 428-432.
LIU Guohai, JIANG Xingke, MEI Congli. Soft sensing of key state variables based on continuous hidden Markov model for fermentation process[J]. Journal of Jiangsu University(Natural Science Eidtion)
, 2011, 32(4): 428-432.