【摘要】 本文针对抗生素发酵生产过程机理复杂,影响因素众多,非线性严重的特点。从几个方面改进了一种启发性自组织的辨识方法—GMDH,并以工业生产的历史批报数据为依据,建立了庆大霉素发酵过程的质量预报模型.仿真结果表明,预报模型能满足工业生产过程的需要.通过质量预报和控制,能完善生产系统的信息管理,提高生产的质量。更多还原

【Abstract】 Antibiotic fermentation processes on industrial production scale, characterized by such complexities as multi-variable and ill-defined mathematical structures, are prone to difficulties in modelling for on-specification and product quality prediction. Grouping Method for Data Handling, GMDH for short, a well-known heuristic self-organizing identification approach for complex systems, modified by the authors in four respects as stated in this paper, was utilized to ease the problem. The prediction model, being verified by field operation data, is expected to be able to up-grade information management and production level of the processes concerned.更多还原