Font Size: a A A

Study Of Ferment Process Control Based On Predictive Control

Posted on:2009-11-04Degree:MasterType:Thesis
Country:ChinaCandidate:Y XiFull Text:PDF
GTID:2121360272956585Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
The Ferment Process has the character of nonlinear, changeless and complexity. Some important Process Variables are unable to measure on line. Traditional Analytic Model is difficult to realize Ferment Process Control and Optimum.Predictive Control correct indetermination influence by Optimize and Feedback, so it is adapt to the complicated environment better than traditional Optimization Control. MPC is method is a new computer control arithmetic which has many merits such as lower model restriction, on-line calculation facility. Predictive Function Control pays more attention to the structure, regard the input as combinations of control variables. So it's computational complexity is less than the common MPC. MPC is used successfully in chemical, metallurgy and petroleum process, but it is used in Ferment Process less. So studying Ferment Process Control with MPC has great meaning.Based on the characters of PFC, combined with Penicillin Ferment characters, study the multi-model PFC with Piecewise-Linearize and the BP Network prediction PFC. Mainly deduce the control variable with the two methods.Apply the multi-model PFC with Piecewise-Linearize to Penicillin Ferment Process. Piecewise-Linearize to attain the Local Models, design the controller for each Local Model. Improve the Weights Device and weight the controllers to attain the General Controller.Apply the BP Network prediction PFC to Penicillin Ferment Process. Study the Feedforward Neural Network prediction structure of the penicillin concentration output equation, obtain the coefficients. The penicillin concentration of penicillin fermentation process is measured by the BP Network modeling methods parameters, CER can be measured online. Use BP to build the Neural Network Controller of PFC,discuss the correction method of the weighted coefficients.The simulation showed the output curve is approximate reference trajectory and has no violent oscillation .The Ferment Process is stationary. In the late, the output of Penicillin is high.
Keywords/Search Tags:Process Control, Predictive Control, Ferment, Neural Network, Penicillin
PDF Full Text Request
Related items