Study on Parameter Optimization, Measurement and Control Technology for CO2 Supercritical Extraction

Author:Li Bing Lin

Supervisor:you wen


Degree Year:2019





Supercritical fluid extraction,which is a new extraction technology,can extract the effective components with high boiling point or heat sensitivity from liquid or solid with high pressure and high density supercritical fluid as extraction agent.Supercritical fluid extraction is a pollution-free extraction technology,which is widely used in food and chemical industry.The study on supercritical fluid extraction has been developed rapidly and got some achievements in many fields,however,there are still many problems need to be further researched and discussed.In the separation process,the solvent extraction performance has been affected by temperature,pressure,flow and other parameters.In this paper,a CO2 supercritical extraction process is taken as the research object to conduct an in-depth study on the modeling,optimization and control of temperature and pressure,which are the two controlled variables in this system.The main research contents and the innovation work in this paper are shown as follows:First,the linear part of the nonlinear temperature-pressure model can be described by non-equidistant grey optimization model.Peng-Robinson equation of state is used to describe the nonlinear part of the temperature-pressure model.Discussing the input/output decoupling condition of the nonlinear temperature-pressure system,a state and output transformation method of this class of system is proposed in this work.It can transform a input-output decoupling system into a disturbance decoupling system to realize the complete decoupling control of temperature and pressure.The decoupling PID controls of the nonlinear temperature-pressure model can be implemented in this work.Second,for the problem on parameters optimization of supercritical fluid extraction,RBF neural network prediction technology has been researched.With the data measured in experiment,a vector,which consists of pressure,temperature,and CO2 flow rate,is constructed to train RBF neural network for predicting extraction rate with different process parameters.Third,using transfer function and state space method to analyze the analytical results of extended state observer in the standardization transformation in linear active disturbance rejection control,using the root locus to analyze the characteristics of transformation results with different parameters,using the Bode diagram to analyze the control effect of PD controller,some conclusions and parameters setting principles are given to provide the theoretical basis for active disturbance rejection controller.Fourth,discussing the methods of the controller disign and parameter turning,an active disturbance rejection controller is proposed.While there is external sinusoidal or chaotic disturbance,temperature PID control can not achieve temperature stability control.Although pressure PID control can achieve pressure stability control,there is a big fluctuation.While there is a wide frequency sinusoidal disturbance,active disturbance rejection controller can achieve stability control for temperature and pressure with the small residual and good disturbance suppression ability.The temperature residual is always less than 0.1℃and decreases with the increase of interference frequency.The performance of the active disturbance rejection controller is better than PID control.Finally,the main content of this dissertation is summarized,and further works are discussed.