Research On Key Technologies Of Intelligent Measurement And Control System For Fermentation Process

Fermentation is the basis of bioengineering as well as modern biology technology and bioengineering industrialization. With the development of the fermentation technology and the expanding of the production scale for fermentation industrial, it is urgent to execute optimal control for fermentation process. The current fermentation control systems (MCU, IPC, PLC, DCS and FCS, etc) lack the intelligent measurement unit which can measure the key biogoical parameters, such as biomass, substrate or product concentrations. The mechanism of fermentation process is very complicated. It has high nonlinearity and time-variant property, and some key biological parameters are difficult to be measured on-line; the architecture of these systems is unit structure and their openness and reliability are poor, which make the industrial application of optimal control algorithms and methods hard to achieve. And they can not meet the need of optimizing control for fermentation process. Therefore, it is important in both theory and application to research the intelligent measurement and control system and its key technologies. Based on the analysis of integration method and key technologies of current measurement and control system for fermentation process, the integration method of intelligent measurement and control system for fermentation process and its key technologies are researched. An integration method of intelligent measurement and control system for fermentation process was proposed; The openness and reliability of the system were discussed; The software architecture of the system was researched; The implementation method of the system was given; The intelligent measurement unit using artificial intelligence technology is integrated in the system and it provides a new approach for the development of measurement and control system for fermentation industry.It is not easy to get good results for the traditional single modeling method because of the complicated mechanism of fermentation process. For solving this problem, a hybrid soft-sensor modeling method using state space model and Suppport Vector Machines (SVM) was proposed. Considering complicated model of fermentation process, the Swarm Energy Conservation Particle Swarm Optimization (SECPSO) algorithm was presented for determination of the simplified mechanism model. The simplified model was taken as the state equations of the hybrid model. The SVM models were adopted to supplement the nonlinear measurement equations of the state space model. The hybrid model can effectively express the main characteristics of the fermentation process and the proposed method is an effective approach to address the problem of on-line accuracy measurement for fermentation process.A Strong Tracking Unscented Kalman Filter (STUKF) with weaken factor was presented according to the thought of Strong Tracking Finite Difference Extend Kalman Filter (STFDEKF). In order to enhance the robustness of STUKF, the adaptive noise estimator was introduced to STUKF and an Adptive STUKF (ASTUKF) algorithm was proposed. The ASTUKF has strong robustness for every initial parameter and the accuracy requirement can also be satisfied. This filter algorithm can be used in the on-line state estimation of high accuracy for fermentation process for its simplicity and good nemerical stability. The ASTUKF algorithm can also be used in different complicated nonlinear system and has the important application valve for practical engineering.The industrial yeast fermentation simulation model was built and the hybrid soft-sensor model was established based on this simulation model. The experiment of state estimation was done based on the hybrid model. The results of the experiment indicate that the hybrid soft-sensor modeling method is effective, and the ASTUKF algorithm is suitable for state estimation for fermentation process. The ASTUKF has strong numerical stability and robust and the accuracy of the filter is guaranteed simultaneously. The penicillin hybrid soft-sensor model was built using the data generated by penicillin simulator. The experiment results prove the compensation effect of the SVM supplement equations to the estimation results in the hybrid soft-sensor model. Through the above experiments, it shows that the proposed hybrid modelilng method and the ASTUKF algorithm for on-line state estimation of fermentation process can enhance the system's robustness effectively to the disturbances of noises and unmodeled errors and improve the accuracy of estimation. It will help to promote the development of the engineering application of intelligent measurement technology for fermentation process to a great extent.The intelligent measurement and system based on PXI bus was integrated for fermentation process. This system was applied in L-lactic acid fermentation process optimization control experiment. It makes the on-line fermentation process optimal control become true and the final product rate of fermentation is increased. It has shown that the proposed integration method of intelligent measurement and control system is feasible and the developed system is the excellent platform for the optimal control of fermentation process. The intelligent measurement and control system is a key system during the practical application of intelligent optimal control algorithms for fermentation process.The system architecture of the proposed intelligent measurement and control system for fermentation process has high openness and reliability, and it provides a new approach and has the guidance meaning for the implement of measurement and control system for fermentation process. The proposed intelligent measurement and system based on PXI bus can achieve the on-line measurement and control of the key biological parameters, and it has wide application prospect in the field of fermentation process control.