| Chlortetracycline(CTC),a kind of tetracyclic broad-spectrum antibiotic,is characterized by bacterial inhibition,promoting animal growth,lower production costs and so on,which has been widely used in medical,agricultural and animal husbandry.During the CTC fermentation process,if the fermenter is invaded by other harmful bacteria,the fermentation broth will be contaminated,and it will influence the growth of beneficial bacteria and the quality of metabolites.Once this occurs,the fermenter needs to be sterilized in time for preventing contamination of other fermenters.When the contamination is serious,it is necessary to discharge the fermentation broth and this will cause a huge waste of raw materials.If the sign of fermentation can be found as early as possible,carrying out sterilization timely will reduce the loss to a minimum.Therefore,it is of great theoretical significance and practical value to do the research of prediction modeling methods for contamination in the CTC fermentation process.In order to predict whether the CTC fermentation process is contaminated effectively,this dissertation studies the soft sensor modeling method based on data-driven principle and the field data of fermentation process.A soft sensor model that predicts the sign of contamination by dissolved oxygen(DO)concentration is presented in the dissertation,the relationship between DO and contamination is studied by ensemble Gauss process regression model based on sample partition,and the validity of the method is verified through the field data.Another soft sensor model is proposed in the dissertation to research the relevance of contamination and the viscosity of the broth,which is an adaptive soft sensor model based on just-in-time learning and ensemble learning,and its validity is also verified through the real field data.The experimental results show that the proposed soft sensor modeling methods can judge whether the CTC fermentation process is contaminated,and predict the contamination in time.The prediction methods of contamination proposed in the dissertation can also be applied to other biological fermentation process. |
