Design of an optimal PI controller for a non-linear process using genetic algorithm. (1st October 2010)

• Record Type: Journal Article
• Authors: Rathikarani, D.; Sivakumar, D.
• Is Part Of: International journal of automation and control. Volume 4:Number 4(2010)
• Page Start: 480
• Page End: 503
• Publication Date: 2010-10-01
• Subjects: LSE -- least squares estimation -- stabilising sets -- GAs -- genetic algorithms -- optimisation -- nonlinear processes -- optimal control -- PI control -- controller design; Automation -- Periodicals; Control theory -- Periodicals; Automatic control -- Periodicals; Robots -- Periodicals; Bionics -- Periodicals; Intelligent control systems -- Periodicals
• Languages: English
• ISSNs: 1740-7516

Abstract

Many industrial processes have to satisfy different criteria to achieve better quality product and performance. These multiple performance criteria need to be optimised simultaneously. However, a suitable optimal solution meeting the entire criteria can hardly be found since these criteria are conflicting. Compared to conventional optimisation techniques, genetic algorithms (GAs) are well suited to solve optimisation problems that involve multiple performance criteria. This paper presents the application of GA to optimise the PI controller parameters for a non-linear process. Least square estimation (LSE) method is used to estimate the parameters of the process. All the PI controllers which stabilise the given plant in the specified ranges are designed using the estimated plant parameters. GA searches the optimal controller parameters within this stabilising set. Experimental results are provided, showing the effectiveness of the multiple performance criteria tuned controller for a nonlinear process.