PHASE-CONTROL OF FED-BATCH CULTURE FOR ALPHA-AMYLASE PRODUCTION BASED ON CULTURE PHASE IDENTIFICATION USING FUZZY INFERENCE

An on-line control system for fed-batch culture was developed. The ''phase control'' system identified different culture phases by using fuzzy set theory. With this system, it was possible to identify culture phases in a fermentor based on current culture data and to automate control of the substrate feed rate. The fuzzy inference system consisted of membership functions for state and control variables, production rules prepared from information provided by experienced operators and a fuzzy inference engine. Culture phases were identified by calculating the adaptability of current culture data to the production rules that described each phase. By following the changes in adaptability, transitions between culture phases could be successfully identified. The substrate feed rate was decided and controlled according to the inferred description of each culture phase. The system was tested by incorporating it into a fed-batch culture for alpha-amylase production by Bacillus amyloliquefaciens. The experimental results showed that the system was able both to identify a change in the microbial state by recognizing culture phase transitions and to use the inferred results of the culture phases to control the feed rate of the substrate.