Research Of Expert System For Biomass Soft-Sensor Modeling In Fermentation Process

Animalcule fermentation engineering is the foundation of bioengineering and modern biology technology as well as industrialization. With the development of fermentation engineering, automatic control technology is presented the higher requirements. Biomass is one of the most important variables in fermentation process, and directly influences the control and optimization of fermentation process. However, with the lack of biological sensors, there are no effective on-line measurements for biomass at present. Soft-sensor technique is an effective approach acquiring non-measuring primary variables. Therefore, it's an effective on-line approach to estimate the biomass using soft-sensor technology in fermentation process.Applying expert system in biomass soft-sensor hybrid modeling in fermentation process, establishing knowledge-base provided with prior knowledge and mechanism in fermentation process, the structure of soft-sensor hybrid model can be acquired by reasoning. It is the foundation of biomass soft-sensor hybrid modeling. There are the important theoretical meaning and practical applied value for on-line estimation of biomass in fermentation process. In this paper, on the base of analyzing the development actuality of biomass soft-sensor detailedly, and fermentation process and expert system were researched.Firstly, prior knowledge and mechanism during researching fermentation process were analyzed. Starting with bacterium, culture medium, attention product, reactive condition in fermentation process, and so on, a knowledge classified-to-get method of expert system for biomass soft-sensor modeling in fermentation process was presented, and then knowledge of fenmentation process was classified effectively.Secondly, in research of knowledge representation, according to characteristics of relational parameters, variables, and mechanism models in fermentation process, a knowledge representation of production rules based on database was presented, and layered organization structure of knowledge was introduced. It could express commendably correlative facts and models in fermentation process. And then a detecting redundancy method for knowledge-base in expert system for biomass soft-sensor modeling in fermentation process was proposed.In research of knowledge reasoning, A forward heuristic breadth-first reasoning methord based on match degree was proposed. It could reason complex model structure in the event of few known facts, and definitions and computations of match degree and reliability were introduced.Finally, according to proposed knowledge representation and reasoning approach, the architecture of expert system for biomass soft-sensor modeling in·fermentation process was provided. A prototype of expert system for biomass soft-sensor modeling in fermentation process was established, and experiments for inspecting feasibility and validity of expert system for biomass soft-sensor modeling in fermentation process were presented.Results show that the knowledge representation of production rules based on database can express commendably correlative facts and mechanism models in fermentation process. The proposed detecting redundancy method for knowledge-base in the expert system is effective and feasible, and is convenient for management and maintenance of knowledge-base. The proposed forward heuristic breadth-first reasoning methord based on match degree and layered organization structure of knowledge can reduce effectively reasoning steps, enhance reasoning efficiency of expert system. The fermentation of Nosiheptide is as research object. The experiment result of expert system for biomass soft-sensor modeling in fermentation process is coincident with conclusion in the reference, and higher estimating accuracy of hybrid model can be obtained if adding prior knowledge and mechanism to the hybrid modeling process.