Methodological Research of the Fuzzy Control in Insulin Production

本研究的目的是开发适合酿酒酵母发酵生产胰岛素的模糊控制系统。胰岛素是治疗糖尿病的重要药物，主要生产在国外，例如Novo。丹麦胰岛素有限公司一直使用酵母菌生产胰岛素，但由于酵母菌产生酒精，严重影响细胞生长和生产，生产中的技术难题可想而知。因此，我们需要进行大量的培养实验，以开发模糊控制算法和在线控制系统。为了确保模糊控制系统的有效性，我们尝试将该控制系统应用于酿酒酵母的酒精生产。我们成功地控制了葡萄糖补料，提高了酒精生产的生长速度，缩短了培养时间。我们的模糊控制具有简单的结构…与日本清酒公司开发的常规模糊控制相比，模糊规则和隶属度函数有更多的要求。将该方法应用于固定化细胞酒精生产系统，成功地将葡萄糖浓度控制在适宜的乙醇生产水平。在模糊控制中，人工调整隶属度函数和人工调整产生式规则是一个严重的问题，这是因为我们在程序上花费了大量的精力，严重影响了模糊控制的效率。然后尝试制作了显示系统，显示了导致模糊推理错误的隶属函数。考虑到显示系统的建议，我们修改了隶属度函数。如果修改较小，模糊控制通常会稍有改进。但是，如果大量修改，模糊控制就会受到破坏，因为修改会影响到许多没有考虑到修改效果的规则。我们还研究了蛋白质组分析技术在模糊控制开发中的应用。目前，二维电泳胶图像并不稳定，但我们利用人工神经网络成功地实现了对培养状态的识别。因此，我们将能够利用蛋白质组分析的信息来开发模糊控制系统。较少

The objective of the research is the development of Fuzzy control system, which include the know-how of the superior fermentation technique, and is suitable for the production of Insulin by Saccharomyces cerevisiae. Insulin is important pharmaceuticals for diabetes, and produced mainly abroad, for example, Novo. Co. Ltd in Denmark has been producing Insulin using yeast However, we can easily imagine the severe technical problems in the production, because yeast produce alcohol, which seriously effect on the cell growth and the production. There fore, we need to perform many culture experiment in order to develop the fuzzy control algorithm and the on-line control system.In order to make sure the validity of the fuzzy control system, we tried to use the control system in the alcohol production by Saccharomyces cerevisiae. We succeeded to control the glucose feeding and to increase the growth rate and shorten culture time for the alcohol production. Our fuzzy control has a simple structu … More re of fuzzy rules and membership functions compared with the normal fuzzy control, which has been developed in Japanese sake brewery company. Our approach applied to the immobilized cell system for alcohol production, and we can keep the glucose concentration at suitable level for the ethanol production successfully. In the fuzzy control, we have a severe problem which is the manual tuning of membership functions and manual adjustment of production rules, because we spend a lot of trouble for the procedures and they influence the efficiency of the fuzzy control serious ly. Then we tried to make the display system which shows the membership functions causing the fuzzy reasoning error. Taking into consideration of the suggestion from the display system, we modified the membership function. If the modification is small, the fuzzy control usually improved a little bit. But, if modified largely, the fuzzy control is damaged, because the many rule, which is not considered into the modification effect was effected by the modification. We also investigate the use of the proteome analysis technique for the fuzzy control development. Now, 2-D electrophoresis gel image is not stable but we succeed it to distinguish the culture state with the aid on ANN. Therefore we will be able to use the information from proteome analysis for the development of fuzzy control system. Less