@article{open910,
          volume = {80},
          number = {5},
          author = {P R Patnaik},
            note = {Copyright of this article belongs to Wiley.},
           title = {Neural Optimization of Fed-batch Streptokinase Fermentation in a Non-ideal Bioreactor},
       publisher = {wiley},
            year = {2008},
         journal = {The Canadian Journal of Chemical Engineering},
           pages = {920--926},
        keywords = {streptokinase;fed-batch reactor;non-ideal;neural optimization},
             url = {http://crdd.osdd.net/open/910/},
        abstract = {Microbial fermentations involving two or more kinds of competing cells and operating under realistic conditions are difficult to monitor, model and optimize by model-based methods. They deviate from ideal behavior in two significant aspects: incomplete dispersion in the broth and the influx of disturbances. The approach here has been to optimize the filtered noise and dispersion on-line through neural networks. This method has been applied to the fed-batch production of streptokinase (SK). The culture has two kinds of cells {--} active (or productive) and inactive {--} and their growth is inhibited by the substrate and the primary metabolite (lactic acid). Using simulated data, the fermentation was optimized by a system of three neural networks, updated continually during successive time intervals. Such sequential optimization with dynamic filtering of inflow noise generated better cell growth and SK activity than static optimization and even an ideal fermentation.

Les fermentations microbiennes faisant intervenir deux ou plusieurs sortes de cellules en comp{\'e}tition et se d{\'e}roulant dans des conditions r{\'e}elles, sont difficiles {\`a} surveiller, {\`a} mod{\'e}liser et {\`a} optimiser par des m{\'e}thodes bas{\'e}es sur des mod{\`e}les. De telles fermentations s'{\'e}cartent du comportement id{\'e}al dans deux voies importantes : la dispersion incompl{\`e}te dans le bouillon et la venue de perturbations. Notre approche consiste ici {\`a} optimiser le bruit filtr{\'e} et la dispersion en continu par des r{\'e}seaux neuronaux. Cette m{\'e}thode a {\'e}t{\'e} appliqu{\'e}e {\`a} la production {\`a} alimentation discontinue de streptokinase (SK). La culture comporte deux sortes de cellules {--} actives (ou productives) ou inactives {--} et leur croissance est inhib{\'e}e par le substrat et la m{\'e}tabolite primaire (acide lactique). {\`A} l'aide de donn{\'e}es simul{\'e}es, la fermentation a {\'e}t{\'e} optimis{\'e}e par un syst{\`e}me de trois r{\'e}seaux neuronaux, qui ont {\'e}t{\'e} mis {\`a} jour continuellement {\`a} des intervalles de temps successifs. Une telle optimisation s{\'e}quentielle avec filtrage dynamique du bruit g{\'e}n{\`e}re une meilleure croissance des cellules et activit{\'e} du SK que l'optimisation statique et m{\^e}me la fermentation id{\'e}ale.
}
}