Research Of Biomass Soft-Sensor Expert System Reasoning Machine In Fermentation Process

The microbial fermentation engineering is the foundation of bioengineering & modern biotechnology and its industrialization. Biomass is the important process parameter of fermentation process, and it directly influences control and optimization for fermentation process. Applying soft-sensor technology in the on-line estimation of biomass parameters is an effective way for biomass parameters online measurement. The key to solve the biomass on-line estimation is to construct hybrid model both using the mechanism model and identification model when building the fermentation biomass soft-sensor. Expert system, which using the experts' knowledge and experience to reason and make decision, is one effective way to build the soft-sensor hybrid model. Reasoning machine is the thinking part of expert system and it is the one important part of expert system. Therefore, the research of the biomass soft-sensor expert system reasoning machine has important theoretical significance and actual value to the building of the biomass soft-sensor hybrid model.Based on analyzing the fermentation process and its characteristics, the structure model of fermentation process biomass soft-sensor expert system reasoning machine was built. The definition of matching degree and its calculation was given based on the knowledge representation form of production rule, which combined with the fermentation process characteristics, such as the related parameters and mechanism formulas, etc. Then, the conflict reduction strategy based on matching degree was presented. The uncertainty reasoning method based on the believable degree was proposed using the concept of believable degree. The mixed direct and inverse heuristic reasoning machine of expert system was developed by using the .NET technology.The experiment results show that conflict reduction strategy based on matching degree can solve the problem of choosing biomass soft-sensor process model rules well. The uncertainty reasoning method based on believable degree can improve believable degree of the conclusion. The designed reasoning machine of soft-sensor expert systemcan obtain better biomass soft-sensor model more fast and accurate.